Hue_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "qULp_oix0Fv5"
   },
   "source": [
    "## Preface and TOC\n",
    "This script is meant to classify a given polygone-based file containing footprint data of buildings of Hue City, Vietnam. \n",
    "The buildings will be classified to one of the following classes:\n",
    "\n",
    "- Shophouse `sh`\n",
    "- Single Family `sf`\n",
    "- Multi Family `mf`\n",
    "- Uninhabited Large `ul`\n",
    "- Uninhabited Small `us`\n",
    "\n",
    "**Table of content**\n",
    "\n",
    "\n",
    "1. [Head: In the first section all the variables and functions are defined and the input-data will be preprocessed](#head)\n",
    "\n",
    "    1.1. [Declaration of variables](#vars)\n",
    "    \n",
    "    1.2. [Declaration of functions](#functions)\n",
    "\n",
    "2. [Main Script](#main)\n",
    "\n",
    "    2.1 [First step: In the main-section the basic parameters are derived from the footprints](#first)\n",
    "    \n",
    "    2.1.1. [Addition of training classes: The given training classes are being added to the previously created dataset for external use in Orange's DT-classifier](#add_train)\n",
    "    \n",
    "    2.1.2. [First classification: The first three classes get classified](#class1)\n",
    "\n",
    "    2.2. [Second step: Derivation of the second step's parameters](#second)\n",
    "\n",
    "    2.2.1. [Second classification: The until here unclassified polygons are being classified](#class2)\n",
    "\n",
    "    2.2.2. [Finalization: The new parameters as well as the new classes are being merged to the dataset, created in 2.1.2. and written to HD](#final)\n",
    "\n",
    "3. [Accuracy assessment: Confusion matrices for accuracy measurements](#acc)\n",
    "\n",
    "4. [Statistics: Statistical evaluations](#statistics)\n",
    "\n",
    "    4.1. [Feature counts: Simple display of the counts of single classes](#feature_counts)\n",
    "\n",
    "    4.2. [Ward related statistics: Counting building types in wards](#wards)\n",
    "\n",
    "    4.3. [Boxplots and histograms: Generation of boxplots and histograms for each parameter, devided by classes](#boxplots)\n",
    "\n",
    "    4.4. [Residential Housings: Statistical evaluation of housings per floor area](#resi)\n",
    "\n",
    "    4.5. [Calculation of parameters' mean values and standard deviation per class](#means)\n",
    "\n",
    "---\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a name=\"head\"></a>\n",
    "# Head"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "xOVrdJmf3G51",
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting package metadata (current_repodata.json): done\n",
      "Solving environment: done\n",
      "\n",
      "# All requested packages already installed.\n",
      "\n",
      "Collecting package metadata (current_repodata.json): done\n",
      "Solving environment: done\n",
      "\n",
      "# All requested packages already installed.\n",
      "\n",
      "Collecting package metadata (current_repodata.json): done\n",
      "Solving environment: done\n",
      "\n",
      "# All requested packages already installed.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import platform\n",
    "if platform.system() == 'Windows':\n",
    "    !conda create -y -n Hue_classification\n",
    "    !conda activate Hue_classification\n",
    "\n",
    "###\n",
    "# NOTE: momepy works better when installed from conda-forge\n",
    "# installation via conda in Google Colab requires condacolab installed\n",
    "# -> condacolab by Jaime Rodríguez-Guerra: https://github.com/jaimergp/condacolab\n",
    "# Not neccessary in Jupyter Notebook\n",
    "# !pip install -q condacolab\n",
    "# import condacolab\n",
    "# condacolab.install()\n",
    "###\n",
    "\n",
    "!conda install momepy -y #-n Hue_classification\n",
    "!conda install plotnine -y #-n Hue_classification\n",
    "!conda install dask -y #-n Hue_classification\n",
    "# %reset clears all variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 2875,
     "status": "ok",
     "timestamp": 1623580259676,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "cKvEqs-vn2iQ",
    "outputId": "3d87fe8c-9717-43a4-8da5-ca3424d27503",
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import momepy as mp\n",
    "import geopandas as gpd\n",
    "import pandas as pd\n",
    "import time \n",
    "import numpy as np\n",
    "from plotnine import *\n",
    "import sklearn.metrics as metrics\n",
    "import warnings\n",
    "import os\n",
    "from tqdm import tqdm\n",
    "from collections import Counter"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "f30TerRxP7VW"
   },
   "source": [
    "<a name=\"vars\"></a>\n",
    "## Declaration of variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 546,
     "status": "ok",
     "timestamp": 1623580262575,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "8sCZdKkNP6fd",
    "outputId": "93d3059b-5873-457f-baca-875026cad8a8",
    "tags": []
   },
   "outputs": [],
   "source": [
    "###### Variables ######\n",
    "\n",
    "update_mode = False  # if True, outfile_first will be loaded instead of infile\n",
    "precision = 5  # precision of calculated values\n",
    "do_tessellation = True  # calculate time demanding tessellation\n",
    "do_enclosed_tess = True  # enclosed tessellation makes only sense if the given data is of higher quality\n",
    "do_preprocessing = True  # time demandeng MP-preprocessing\n",
    "minsize = 2  # buildings smaller than this will be cut out by mp.preprocess()\n",
    "gpd.options.use_pygeos = True  # using PYGEOS speeds up certain operations\n",
    "start_time = time.time()  # starting time for time measurement\n",
    "os.chdir('/media/user/Daten/BA_workdir/Jupyter_workdir/')  # sets the working directory\n",
    "# if there is no subdirectory named 'Output' create one\n",
    "if not os.path.exists('./Output'):\n",
    "    os.mkdir('./Output')\n",
    "    os.mkdir('./Output/Daten')\n",
    "    os.mkdir('./Output/Klassifikation')\n",
    "    os.mkdir('./Output/Statistik')\n",
    "    \n",
    "# data input\n",
    "infile = './Input/Hue_buildings_WGS84.geojson'\n",
    "roads_in = './Input/Hue_osm_roads_free.shp'\n",
    "water_in = './Input/Hue_osm_water_free.shp'\n",
    "training_classes_in = \"./Input/Hue_training_classes_no_params.csv\"\n",
    "landuse_in = None  # './Input/Hue_gis_osm_landuse_a_free.shp'\n",
    "townhall_in = None  # './Input/Hue_town_hall.shp'\n",
    "max_density_spot_in = None  # './Input/Daten/Hue_max_density_spot.shp'\n",
    "k_dense_in = None  # './Input/Hue_kernal_density_1km.csv'\n",
    "reference_classes = './Input/Hue_reference_classes_preprocessed.csv'\n",
    "\n",
    "# data output\n",
    "infile_preprocessed = './Input/Hue_buildings_UTM48N_preprocessed.shp' \n",
    "outfile_first = './Output/Daten/Hue_buildings_UTM48N_preprocessed_first_no_classes.shp'\n",
    "data_w_training_classes_out = \"./Output/Daten/Hue_buildings_UTM48N_preprocessed_edited_w_training_classes.shp\"\n",
    "outfile_classified = './Output/Daten/Hue_buildings_UTM48N_classified_full.shp'\n",
    "tess_out = './Output/Daten/Hue_tessellation.shp'\n",
    "tess_encl_out = './Output/Daten/Hue_tessellation_enclosed.shp'\n",
    "outfile_blocks = './Output/Daten/Hue_blocks.shp'\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Un3RxbnGQAbQ"
   },
   "source": [
    "<a name=\"functions\"></a>\n",
    "## Declaration of functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 28310,
     "status": "ok",
     "timestamp": 1623580293953,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "Zvl1lbX3bL-C",
    "outputId": "d9b18797-3d26-4884-b268-21f21532ba5c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Importing data...\n",
      "Reprojecting to EPSG 32648\n",
      "Preprocessing...\n",
      "Loop 1 out of 2.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Identifying changes: 100%|██████████| 109092/109092 [00:05<00:00, 18809.47it/s]\n",
      "Changing geometry: 100%|██████████| 262/262 [00:22<00:00, 11.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loop 2 out of 2.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Identifying changes: 100%|██████████| 108838/108838 [00:05<00:00, 19003.57it/s]\n",
      "Changing geometry: 100%|██████████| 22/22 [00:02<00:00, 10.31it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collapsed features  : 0\n",
      "Split features      : 18\n",
      "Overlapping features: 49\n",
      "67 features with erroneus geometry were dropped from the GeoDataFrame.\n",
      "0 duplicated features were dropped.\n"
     ]
    }
   ],
   "source": [
    "###### Functions ######\n",
    "\n",
    "def reproject(shp):\n",
    "    \"\"\"\n",
    "    Converts shp to metric CRS if neccesary    \n",
    "    \"\"\"\n",
    "    if shp.crs != 'EPSG:32648':\n",
    "        print(\"Reprojecting to EPSG 32648\")\n",
    "        shp = shp.to_crs(\"EPSG:32648\")\n",
    "    return shp\n",
    "\n",
    "\n",
    "def msize(gdf, radius, ignore_current=True):\n",
    "    \"\"\"\n",
    "    Calculates the mean footprint area of surrounding buildings within a given \n",
    "    range, excluding the current buildings area.\n",
    "    Adds 1 new column to the given GeoDataFrame.\n",
    "    \"\"\"\n",
    "\n",
    "    for i, row in tqdm(gdf.iterrows(), total=gdf.shape[0]):\n",
    "        idx = data.sindex.query(row.geometry.buffer(radius), predicate='contains')\n",
    "        if ignore_current:\n",
    "            idx = np.delete(idx, np.where(idx == i))\n",
    "        if len(idx) == 0:\n",
    "            gdf.loc[i, 'msize_' + str(radius)] = 0\n",
    "            continue\n",
    "        else:\n",
    "            gdf.loc[i, 'msize_' + str(radius)] = sum(gdf.iloc[idx].area) / len(idx)\n",
    "    return gdf\n",
    "\n",
    "\n",
    "def mabr(gdf):\n",
    "    \"\"\"\n",
    "    Calculates area, width and length for each opjects \n",
    "    minimum area bounding rectangle in the given gdf.\n",
    "    Adds 3 new columns to the given GeoDataFrame.\n",
    "    \"\"\"\n",
    "    for i, row in tqdm(gdf.iterrows(), total=gdf.shape[0]):\n",
    "        mabr = row.geometry.minimum_rotated_rectangle\n",
    "        gdf.loc[i, 'area_mabr'] = mabr.area\n",
    "        gdf.loc[i, 'width'] = mabr.minimum_clearance\n",
    "        gdf.loc[i, 'length'] = (mabr.length - (2 * mabr.minimum_clearance)) / 2\n",
    "\n",
    "    return gdf\n",
    "\n",
    "\n",
    "def mean_orientation(gdf, radius, ignore_current=True):\n",
    "    \"\"\"\n",
    "    Calculates building's orientations. Then, calculates mean deviation\n",
    "    of each building from surrounding buildings within the given range.\n",
    "    Adds 3 new columns to the given GeoDataFrame.\n",
    "    \"\"\"\n",
    "    if 'orient' not in gdf:\n",
    "        print(\"\\n\")\n",
    "        gdf['orient'] = mp.Orientation(gdf, verbose=True).series\n",
    "\n",
    "    for i, row in tqdm(gdf.iterrows(), total=gdf.shape[0]):\n",
    "        idx_orient = gdf.sindex.query(row.geometry.buffer(radius),\n",
    "                                      predicate='intersects')\n",
    "        if ignore_current:\n",
    "            idx_orient = np.delete(idx_orient,\n",
    "                                   np.where(idx_orient == i))\n",
    "        sum_orient = sum(gdf.iloc[idx_orient].orient)\n",
    "        try:\n",
    "            gdf.loc[i, 'orient_' + str(radius)] = sum_orient / len(idx_orient)\n",
    "        except ZeroDivisionError:\n",
    "            gdf.loc[i, 'orient_' + str(radius)] = row.orient\n",
    "    gdf['orient_dev'] = abs(gdf.orient - gdf['orient_'+ str(radius)])\n",
    "\n",
    "    return gdf\n",
    "\n",
    "\n",
    "print(\"Importing data...\")\n",
    "if update_mode:\n",
    "    data = reproject(gpd.read_file(outfile_first))\n",
    "    data.FID = data.FID.astype(np.int64)\n",
    "else:\n",
    "    data = reproject(gpd.read_file(infile))\n",
    "\n",
    "####### Preprocessings and preparations ######\n",
    "\n",
    "if 'FID' not in data:\n",
    "    data['FID'] = mp.unique_id(data)\n",
    "\n",
    "if do_preprocessing and not update_mode:\n",
    "    print(\"Preprocessing...\")\n",
    "    data = mp.preprocess(data, size=minsize)\n",
    "    defect = mp.CheckTessellationInput(data)\n",
    "    data = data.drop(defect.split.append(defect.overlap.append(defect.collapse)).index)\n",
    "    n_drop = data.shape[0] - data.FID.drop_duplicates().shape[0]\n",
    "    data.to_file(infile_preprocessed, driver='GeoJSON', encoding='utf-8')\n",
    "    print(\"{} features with erroneus geometry were dropped from the GeoDataFrame.\".format(\n",
    "        len(defect.split.append(defect.overlap.append(defect.collapse)))))\n",
    "    print(\"{} duplicated features were dropped.\".format(data.shape[0] - data.FID.drop_duplicates().shape[0]))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "FvW3fPXVz84o"
   },
   "source": [
    "<a name=\"main\"></a>\n",
    "# Main script\n",
    "\n",
    "---\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "qafjWOiBc7iY"
   },
   "source": [
    "<a name=\"first\"></a>\n",
    "## First step: generation of the base parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "cellView": "code",
    "colab": {
     "background_save": true
    },
    "id": "DOCbCTI26gZ-"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating areas...\n",
      "Calculating corners...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/miniconda3/lib/python3.9/site-packages/momepy/shape.py:795: RuntimeWarning: invalid value encountered in arccos\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/momepy/shape.py:794: RuntimeWarning: invalid value encountered in double_scalars\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating perimeters...\n",
      "Calculating rectangularity values...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/108749 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating perimeter per area...\n",
      "Calculating shape index...\n",
      "Calculating length, width and area of MABRs...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 108749/108749 [06:18<00:00, 287.03it/s]\n",
      "  0%|          | 205/108749 [00:00<00:52, 2049.43it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating elongations...\n",
      "Calculating orientations...\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 108749/108749 [00:53<00:00, 2019.61it/s]\n",
      "100%|██████████| 108749/108749 [25:36<00:00, 70.78it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating shared walls...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/108749 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating proximity...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 108749/108749 [05:27<00:00, 332.33it/s]\n",
      "  0%|          | 0/108749 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counting buildings within 50, 100, 200 m and 1 ha range...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 108749/108749 [11:11<00:00, 161.99it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Preparing roads...\n",
      "Reprojecting to EPSG 32648\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/108749 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating distances to roads...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 108749/108749 [06:31<00:00, 277.87it/s]\n",
      "  0%|          | 0/108749 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counting roads within 100 and 250 m range...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 108749/108749 [04:34<00:00, 395.54it/s]\n",
      "  0%|          | 0/108749 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating mean area of buildings within 50 m and 100 m range\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 108749/108749 [26:07<00:00, 69.37it/s]\n",
      "100%|██████████| 108749/108749 [27:52<00:00, 65.04it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating tessellation layer...\n",
      "Reprojecting to EPSG 32648\n",
      "Reprojecting to EPSG 32648\n",
      "Writing tessellation results to file...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Adding \"tess_area\" for single inbound buildings:   0%|          | 0/55 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Adding tessellation related information: tess_area and CAR...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Adding \"tess_area\" for single inbound buildings: 100%|██████████| 55/55 [00:00<00:00, 61.55it/s]\n",
      "Adding \"tess_area\": 100%|██████████| 108749/108749 [02:55<00:00, 619.44it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generating blocks...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Adding \"bID\" to dataset: 100%|██████████| 108749/108749 [02:39<00:00, 681.05it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Calculating building's mean sizes and built up ratio within blocks...\n",
      "Writing output file...\n",
      "\n",
      "Process finished. It took 149 minutes.\n",
      "Output written to './Output/Daten/Hue_buildings_UTM48N_preprocessed_first_no_classes.shp.\n"
     ]
    }
   ],
   "source": [
    "###### Main script ######\n",
    "print(\"Calculating areas...\")\n",
    "data['area'] = data.geometry.area\n",
    "\n",
    "print(\"Calculating corners...\")\n",
    "data['corners'] = mp.Corners(data, verbose=False).series\n",
    "\n",
    "print(\"Calculating perimeters...\")\n",
    "data['perimeter'] = mp.Perimeter(data).series\n",
    "\n",
    "print(\"Calculating rectangularity values...\")\n",
    "data['rect'] = mp.Rectangularity(data, areas='area').series\n",
    "\n",
    "print(\"Calculating perimeter per area...\")\n",
    "data['shp_ratio'] = data.area / data.perimeter\n",
    "\n",
    "print(\"Calculating shape index...\")\n",
    "data['shp_index'] = data.perimeter / (2 * np.sqrt(np.pi * data.area))\n",
    "\n",
    "print(\"Calculating length, width and area of MABRs...\")\n",
    "data = mabr(data)\n",
    "\n",
    "print(\"Calculating elongations...\")\n",
    "data['elongation'] = data.width / data.length\n",
    "\n",
    "print(\"Calculating orientations...\")\n",
    "data = mean_orientation(data, 100)\n",
    "\n",
    "print(\"Calculating shared walls...\")\n",
    "data['shrd_walls'] = abs(np.round(mp.SharedWalls(data).series, 0))\n",
    "\n",
    "print(\"Evaluating proximity...\")\n",
    "for i, row in tqdm(data.iterrows(),\n",
    "                   total=data.shape[0]):\n",
    "    if row['shrd_walls'] == 0:\n",
    "        data.loc[i, 'proximity'] = 'detached'\n",
    "    elif row['shrd_walls'] <= int(row['length']):\n",
    "        data.loc[i, 'proximity'] = 'semi-detached'\n",
    "    else:\n",
    "        data.loc[i, 'proximity'] = 'attached'\n",
    "\n",
    "print(\"Counting buildings within 50, 100, 200 m and 1 ha range...\")\n",
    "for i, row in tqdm(data.iterrows(),\n",
    "                   total=data.shape[0]):\n",
    "    data.loc[i, 'n_in_50m'] = len(data.sindex.query(row.geometry.buffer(50),\n",
    "                                                    predicate='intersects'))\n",
    "\n",
    "    data.loc[i, 'n_in_100m'] = len(data.sindex.query(row.geometry.buffer(100),\n",
    "                                                     predicate='intersects'))\n",
    "\n",
    "    data.loc[i, 'n_in_200m'] = len(data.sindex.query(row.geometry.buffer(200),\n",
    "                                                     predicate='intersects'))\n",
    "\n",
    "    data.loc[i, 'n_in_ha'] = len(data.sindex.query(row.geometry.buffer(\n",
    "        np.sqrt(10000 / (2 * np.pi))),\n",
    "        predicate='intersects'))\n",
    "\n",
    "# TODO: Raus!\n",
    "# Add landuse information based on OSM data\n",
    "if landuse_in is not None:\n",
    "    print(\"Adding OSM-landuse information...\")\n",
    "    landuse = reproject(gpd.read_file(landuse_in))\n",
    "    index_within = []\n",
    "    data['landuse'] = 'unknown'\n",
    "    for i in tqdm(range(len(landuse))):\n",
    "        index_within = data.geometry.within(landuse.geometry.iloc[i])\n",
    "        index_within = index_within[index_within]\n",
    "        for j in range(len(index_within)):\n",
    "            data.loc[index_within.index[j], 'landuse'] = landuse.loc[i, 'fclass']\n",
    "\n",
    "# TODO: Raus!\n",
    "# Add kernal density information to each feature\n",
    "# if k_dense_in is not None:\n",
    "#     print(\"Adding kernal density information...\")\n",
    "#     k_dense = gpd.read_file(k_dense_in)\n",
    "#     k_dense.k_dens_1km = k_dense.k_dens_1km.astype(float)\n",
    "#     k_dense.FID = k_dense.FID.astype(np.int64)\n",
    "#     data = data.merge(k_dense[['FID', 'k_dens_1km']], on='FID')\n",
    "\n",
    "# Calculate distances to roads\n",
    "if roads_in is not None:\n",
    "    print(\"Preparing roads...\")\n",
    "    roads = reproject(gpd.read_file(roads_in))\n",
    "    roads = mp.extend_lines(roads, 0.25)\n",
    "    roads = mp.close_gaps(roads, 0.25)\n",
    "    roads = mp.remove_false_nodes(roads)\n",
    "    roads_union = roads.unary_union  # unifies all roads\n",
    "    print(\"Calculating distances to roads...\")\n",
    "    for i, row in tqdm(data.iterrows(),\n",
    "                       total=data.shape[0]):\n",
    "        data.loc[i, 'to_roads'] = row.geometry.boundary.distance(roads_union)\n",
    "\n",
    "# Roads in range\n",
    "print(\"Counting roads within 100 and 250 m range...\")\n",
    "for i, row in tqdm(data.iterrows(),\n",
    "                   total=data.shape[0]):\n",
    "    data.loc[i, 'roads_100'] = len(roads.geometry.sindex.query(\n",
    "        row.geometry.buffer(100), predicate='intersects'))\n",
    "    data.loc[i, 'roads_250'] = len(roads.geometry.sindex.query(\n",
    "        row.geometry.buffer(250), predicate='intersects'))\n",
    "\n",
    "# TODO: Raus!\n",
    "# Caculate distances to to town hall\n",
    "if townhall_in is not None:\n",
    "    print(\"Calculating distances to town hall...\")\n",
    "    townhall = reproject(gpd.read_file(townhall_in))\n",
    "    data['to_hall'] = data.geometry.distance(townhall.loc[0, 'geometry'], align=False)\n",
    "\n",
    "# TODO: Raus!\n",
    "if max_density_spot_in is not None:\n",
    "    print(\"Calculating distances to highest density spot...\")\n",
    "    max_density_spot = reproject(gpd.read_file(max_density_spot_in))\n",
    "    data['to_centre'] = data.geometry.distance(max_density_spot.loc[0, 'geometry'], align=False)\n",
    "\n",
    "# Mean size of buildings in a given range \n",
    "# this is very time demanding\n",
    "print(\"Calculating mean area of buildings within 50 m and 100 m range\")\n",
    "data = msize(data, 50)\n",
    "data = msize(data, 100)\n",
    "\n",
    "if do_tessellation:\n",
    "    print(\"Calculating tessellation layer...\")\n",
    "    if not do_enclosed_tess:\n",
    "        limit = mp.buffered_limit(data)\n",
    "        tessellation = mp.Tessellation(data,\n",
    "                                       unique_id='FID',\n",
    "                                       limit=limit,\n",
    "                                       verbose=True\n",
    "                                       ).tessellation\n",
    "        tessellation['tess_area'] = mp.Area(tessellation).series\n",
    "        print(\"Writing tessellation results to file...\")\n",
    "        tessellation.to_file(tess_out)\n",
    "\n",
    "    elif do_enclosed_tess and roads_in is not None and water_in is not None:\n",
    "        # if waters an roads are present do enclosed tessellation:\n",
    "        # enclosed tessellation needs roads and water information for the \n",
    "        # creation of boundaries. May cause problems when intersecting with buildings.\n",
    "        \n",
    "        roads = reproject(gpd.read_file(roads_in))\n",
    "        roads = mp.extend_lines(roads, 0.25)\n",
    "        roads = mp.close_gaps(roads, 0.25)\n",
    "        roads = mp.remove_false_nodes(roads)\n",
    "        water = reproject(gpd.read_file(water_in))\n",
    "        enclosures = mp.enclosures(roads, limit=gpd.GeoSeries(\n",
    "            [data.unary_union.convex_hull.buffer(10)]),\n",
    "                                   additional_barriers=[water.boundary])\n",
    "        tessellation = mp.Tessellation(data,\n",
    "                                       unique_id='FID',\n",
    "                                       enclosures=enclosures,\n",
    "                                       verbose=True\n",
    "                                       ).tessellation\n",
    "        tessellation['tess_area'] = tessellation.area\n",
    "        print(\"Writing tessellation results to file...\")\n",
    "        tessellation.to_file(tess_encl_out)\n",
    "\n",
    "    print(\"Adding tessellation related information: tess_area and CAR...\")\n",
    "    # in rare cases where only one building is situated within a bounded area\n",
    "    # this building won't be envolved in the tessellation. Thus the following \n",
    "    # statement hands the area of the untessellated cell to the building\n",
    "    \n",
    "    FID_not_tessellated = list(\n",
    "        data[data.FID.isin(tessellation[tessellation.FID.isna() == False].FID\n",
    "        ) == False].FID)\n",
    "    if len(FID_not_tessellated) > 0:\n",
    "        for i in tqdm(FID_not_tessellated,\n",
    "                      desc='Adding \"tess_area\" for single inbound buildings'):\n",
    "            tessellation.loc[\n",
    "                tessellation.sindex.query(\n",
    "                    data.loc[data[data.FID == i].index[0]].geometry,\n",
    "                    predicate='within')[0], 'FID'] = i\n",
    "    for i, row in tqdm(data.iterrows(), total=data.shape[0],\n",
    "                       desc='Adding \"tess_area\"'):\n",
    "        data.loc[i, 'tess_area'] = tessellation.loc[tessellation.FID == row.FID, \n",
    "                                                    'tess_area'].values\n",
    "    data['CAR'] = data.area / data.tess_area\n",
    "\n",
    "print(\"Generating blocks...\")\n",
    "blocks = mp.Blocks(tessellation=tessellation,\n",
    "                   edges=roads,\n",
    "                   buildings=data, \n",
    "                   id_name = 'bID',\n",
    "                   unique_id = 'FID')\n",
    "\n",
    "blocks.blocks.to_file(outfile_blocks, encoding='utf-8')\n",
    "\n",
    "for i, row in tqdm(data.iterrows(), total=data.shape[0], desc='Adding \"bID\" to dataset'):\n",
    "    data.loc[i, 'bID'] = blocks.blocks.sindex.query(data.loc[i].geometry,\n",
    "                                                    predicate='within')[0]\n",
    "\n",
    "print(\"Calculating building's mean sizes and built up ratio within blocks...\")\n",
    "data['b_msize'] = data.area\n",
    "data['b_builtup'] = 0\n",
    "data['b_medsize'] = data.area\n",
    "bID_set = list(set(data.bID))\n",
    "for i in range(len(bID_set)):\n",
    "    buildings_in_block = data.loc[data.bID == bID_set[i]]\n",
    "    data.loc[data.FID.isin(buildings_in_block.FID), \n",
    "             'b_msize'] = buildings_in_block.area.sum() / len(buildings_in_block)\n",
    "    data.loc[data.FID.isin(buildings_in_block.FID), \n",
    "             'b_medsize'] = buildings_in_block.area.median()\n",
    "    data.loc[data.FID.isin(buildings_in_block.FID), \n",
    "             'b_builtup'] = buildings_in_block.area.sum() / row.geometry.area\n",
    "\n",
    "# rounds all numerical values to the given precision\n",
    "data = np.round(data, precision)\n",
    "\n",
    "print('Writing output file...')\n",
    "data.to_file(outfile_first, encoding='utf-8')\n",
    "print(\"\\nProcess finished. It took {time} minutes.\\n\"\n",
    "      \"Output written to '{fname}.\".format(time=round((time.time() - start_time) / 60), fname=outfile_first))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "z026X_oBe4KQ"
   },
   "source": [
    "<a name=\"add_train\"></a>\n",
    "### Addition of taining classes\n",
    "Here, the training classes are being applied to the `data` GeoDataFrame. The output copy should only contain features with defined class in the `class_name` attribute. The `FID` attribute serves as identifier. The result is meant to be used during the training of the classifier. \n",
    "\n",
    "Inputs:\n",
    "\n",
    "`Hue_training_classes_no_params.csv` file containing the training classes\n",
    "\n",
    "`outfile_first` should be the same as `data` but is read from the hd just in case `data` doesn't exist\n",
    "\n",
    "Outputs: \n",
    "\n",
    "`Hue_buildings_UTM48N_preprocessed_edited_w_training_classes.shp`\n",
    "\n",
    "`Hue_buildings_UTM48N_preprocessed_edited_w_training_classes.shp.csv`\n",
    "\n",
    "`Hue_training_classes_w_params.csv`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "pGhucYx-esDh"
   },
   "outputs": [],
   "source": [
    "###### Add classes ######\n",
    "\n",
    "classes = gpd.read_file(training_classes_in)\n",
    "classes.FID = classes.FID.astype(np.int64)\n",
    "data_w_training_classes = gpd.read_file(outfile_first)\n",
    "data_w_training_classes.FID = data_w_training_classes.FID.astype(np.int64)\n",
    "for i, row in classes.iterrows():\n",
    "    data_w_training_classes.loc[data_w_training_classes.FID == row.FID,\n",
    "                                'class_name'] = row['class_name']\n",
    "    # data_w_training_classes.loc[data_w_training_classes.FID == row.FID, 'class_info'] = row['class_info']\n",
    "\n",
    "# write only features with set class_name as csv\n",
    "data_w_training_classes[data_w_training_classes.class_name.isnull() == False].to_csv(\n",
    "    \"./Output/Klassifikation/Hue_1st_step_training_classes_w_params.csv\")\n",
    "\n",
    "# write the whole data table as shapefile and csv\n",
    "data_w_training_classes.to_file(data_w_training_classes_out, encoding='utf-8')\n",
    "data_w_training_classes.loc[data_w_training_classes.class_name.isnull() == False, 'class_name']\n",
    "data_w_training_classes.to_csv(data_w_training_classes_out+\".csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "9Y_QqwfKIEVp"
   },
   "source": [
    "<a name=\"class1\"></a>\n",
    "### First classification\n",
    "\n",
    "The new dataset `data_w_classes` is created here. It contains all the information of `data` (parameters). \n",
    "In this first step the classes `us`, `sh` and `sf` are beeing added. The thresholds where calculated with help of the Orange decision tree classification.\n",
    "\n",
    "Input:\n",
    "\n",
    "`outfile_first`-defined file containing all attributes\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "Ha-pxUC7iqT8"
   },
   "outputs": [],
   "source": [
    "data_w_classes = gpd.read_file(outfile_first)\n",
    "data_w_classes.FID = data_w_classes.FID.astype(np.int64)\n",
    "\n",
    "data_w_classes.loc[data_w_classes.area <= 30, 'class_name'] = 'us'\n",
    "\n",
    "data_w_classes.loc[ (data_w_classes.class_name != 'us') &\n",
    "                    (\n",
    "                        (data_w_classes.area > 30) &\n",
    "                        (data_w_classes.area_mabr <= 141.80) &\n",
    "                        (data_w_classes.elongation <= 0.17) &\n",
    "                        (data_w_classes.b_msize <= 120.15)\n",
    "                    ),\n",
    "                    'class_name'] = 'sh'\n",
    "\n",
    "data_w_classes.loc[ (data_w_classes.class_name != 'us') &\n",
    "                    (data_w_classes.class_name != 'sh') &\n",
    "                    (\n",
    "                        (data_w_classes.area > 30) &\n",
    "                        (data_w_classes.area_mabr <= 141.80) &\n",
    "                        (data_w_classes.elongation > 0.17)\n",
    "                    ),\n",
    "                    'class_name'] = 'sf'\n",
    "\n",
    "data_w_classes.loc[data_w_classes.class_name.isnull() == True, 'class_name'] = 'mf_ul'\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gkeAwUg6b_M9"
   },
   "source": [
    "<a name=\"second\"></a>\n",
    "## Second step: generation of further parameters\n",
    "After `sh`, `us` and `sf` have been classified, a subset containing only the potential mf and ul classes is generated here. The following will be done:\n",
    "\n",
    "*   Generate subset `data_mful` \n",
    "*   Calculate additional parameters based on `data_w_classes` from the first step\n",
    "*   Add training class names\n",
    "*   Save `./Output/Daten/Hue_buildings_UTM48N_preprocessed_classes_2nd_step.csv` containing new subset\n",
    "*   Save `./Output/Klassifikation/Hue_2nd_step_training_classes_w_params.csv` containing second steps training classes w/ params\n",
    "*   Save `./Output/Klassifikation/Hue_reference_classes_preprocessed_mful_w_params.csv` containing only `mf` and `ul` reference classes with params\n",
    "\n",
    "Note: These operations can be done using the index because `data_mful` inherits it from `data_w_classes`. \n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 281,
     "referenced_widgets": [
      "201a57ace6c14df68f078c6db7411c2e",
      "d524c1c6f0aa45a787571ddfde83f68a",
      "d2ff2ee31b1b4f19bccb592657fbafba",
      "7c509b8e62224a36af1311f3838a82bf",
      "046afab5c04643cfba91480444490605",
      "d0802212b19043c3be7b4d9235dcc495",
      "8f1f870dad824d059540b879488a4ae5",
      "1c2af11f863a4e89b769f13c24ccdcbc",
      "f58106c95326412db9edf3eefbc89140",
      "59442c7e7a4f4ef4928e1b3f7cd71843",
      "16d1d3418e3a47e4aef4ff040c1f0a64",
      "6ac3313a05884052b86ce3dd001ff160",
      "166973b412084481849fc6b37c431494",
      "b6bcc204e49e486ba2567c556069bd91",
      "360d95d8a9114dc3a2117da84a5f81a2",
      "47053e3b47d74f5ea758c34e3a07a76b",
      "246d31140fd74d7ab95d22fc2b276987",
      "6a9f009e7a9f4019a51dd6ad0adf9b9d",
      "552afccb92824ee99784496a51c89561",
      "356f0e34ef7d446992147c1450868b0c",
      "512cc211d14548b5b90adce91b7a0f5e",
      "02248b97e95e471896dec3c2c2445c2c",
      "eae35f5a58894102b5e0d4d4efbf8d66",
      "bb0c7671947e424e884441b31983fd2b",
      "e14e97bc5c15410385ffcc93c67ca4f1",
      "054ccda05e8248bc97004e0950053b3a",
      "65b11cea931f425eb0ee046ed35ccef0",
      "c7f2dc2a44e6440cb282be20c57d3017",
      "624d5c75304d4db3923b438a33774434",
      "3280f822d845445abb1f49788ce46b5a",
      "aae6bd861fb54b3ebef1e5acdde347c1",
      "2c7f4be35c9143459746b4d16a529381"
     ]
    },
    "executionInfo": {
     "elapsed": 282410,
     "status": "ok",
     "timestamp": 1623584366946,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "U5bLbv21v4pS",
    "outputId": "34a4e9f6-9125-4bfd-c67d-08667aa0dc9e"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 1/5085 [00:00<14:25,  5.87it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Retrieving most common class within 50 m range...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5085/5085 [00:56<00:00, 89.46it/s]\n",
      "  0%|          | 2/5085 [00:00<04:33, 18.55it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Retrieving nearest building's class...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5085/5085 [01:10<00:00, 71.82it/s]\n",
      "  0%|          | 3/1102 [00:00<00:41, 26.56it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Retrieving most common class and share of mf_ul classes within blocks...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1102/1102 [00:41<00:00, 26.75it/s]\n",
      "  0%|          | 23/5085 [00:00<00:22, 229.64it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counting mf and ul classes within 50 m range...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5085/5085 [00:04<00:00, 1235.29it/s]\n"
     ]
    }
   ],
   "source": [
    "data_w_classes_no_us = data_w_classes.copy()\n",
    "# data_mful = data_w_classes[data_w_classes.class_name.isnull() == True]\n",
    "data_mful = gpd.GeoDataFrame(data_w_classes[data_w_classes.class_name == 'mf_ul'])\n",
    "# data_w_classes[(data_w_classes.class_name != 'us') &\n",
    "#                            (data_w_classes.class_name != 'sf') &\n",
    "#                            (data_w_classes.class_name != 'sh')]    \n",
    "\n",
    "# data_mful = mp.preprocess(data_mful, size=20) # unifies touching features\n",
    "# data_mful = data_mful.drop(columns='class_name', axis=1)\n",
    "\n",
    "# with or without centroid makes a big difference here\n",
    "print(\"Retrieving most common class within 50 m range...\")\n",
    "for i, row in tqdm(data_mful.iterrows(), total=data_mful.shape[0]):\n",
    "    found_classes = data_w_classes_no_us.iloc[data_w_classes_no_us.sindex.query(\n",
    "        row.geometry.buffer(50),\n",
    "        predicate='intersects')\n",
    "        ].class_name\n",
    "    # found_classes = found_classes.drop(index=i) # remove subject\n",
    "    if len(found_classes) > 0:\n",
    "        data_mful.loc[i, 'class_50'] = Counter(found_classes).most_common(1)[0][0]\n",
    "    else:\n",
    "        data_mful.loc[i, 'class_50'] = 'mf_ul'\n",
    "\n",
    "print(\"Retrieving nearest building's class...\")\n",
    "for i, row in tqdm(data_mful.iterrows(), total=data_mful.shape[0]):\n",
    "    # TODO\n",
    "    with warnings.catch_warnings():  # suppresses overflow warning in the distance process\n",
    "       warnings.simplefilter(action='ignore', category=Warning) \n",
    "       nearest = data_w_classes_no_us.iloc[data_w_classes_no_us.sindex.query(\n",
    "            row.geometry.buffer(50),\n",
    "            predicate='intersects')].geometry.distance(row.geometry)\n",
    "       nearest = nearest.drop(index=i)  # remove current object from lcolab_classesist\n",
    "    if len(nearest) == 0:\n",
    "        continue\n",
    "    # if there are more than one dist = 0 classes \n",
    "    # and at least one of them is mf_ul, set mf_ul as nearest\n",
    "    else: \n",
    "        if len(nearest[nearest == 0]) > 1 and 'mf_ul' in data_w_classes_no_us.loc[\n",
    "                nearest[nearest == 0].index, 'class_name'].values:\n",
    "            data_mful.loc[i, 'nearest'] = 'mf_ul'\n",
    "            continue\n",
    "        nearest = data_w_classes_no_us.loc[nearest[nearest == min(nearest)\n",
    "                                                   ].index[0], 'class_name']\n",
    "        if type(nearest) is not str:\n",
    "            data_mful.loc[i, 'nearest'] = 'mf_ul'\n",
    "        else: # if nearest \n",
    "            data_mful.loc[i, 'nearest'] = nearest\n",
    "\n",
    "bID_set = list(set(data_w_classes_no_us.bID))\n",
    "print(\"Retrieving most common class and share of mf_ul classes within blocks...\")\n",
    "for i in tqdm(bID_set):\n",
    "    buildings_in_block = data_w_classes_no_us[data_w_classes_no_us.bID == i]\n",
    "    # index = data_w_classes_no_us.loc[buildings_in_block.index].index\n",
    "    data_mful.loc[data_mful.FID.isin(buildings_in_block.FID), 'b_class'] = \\\n",
    "        Counter(buildings_in_block.class_name).most_common(1)[0][0]\n",
    "\n",
    "    b_mful_rat = np.round(\n",
    "        len(buildings_in_block[buildings_in_block.class_name == 'mf_ul']) / \\\n",
    "        len(data[data.bID == buildings_in_block.iloc[0].bID]), precision)\n",
    "    data_mful.loc[data_mful.FID.isin(buildings_in_block.FID), 'b_mful_rat'] = b_mful_rat\n",
    "\n",
    "print(\"Counting mf and ul classes within 50 m range...\")\n",
    "for i, row in tqdm(data_mful.iterrows(), total=data_mful.shape[0]):\n",
    "    mful_idx = data_mful.sindex.query(row.geometry.buffer(50), predicate='intersects')\n",
    "    mful_idx = np.delete(mful_idx, np.where(mful_idx == i))\n",
    "    data_mful.loc[i, 'mful_50m'] = len(mful_idx)\n",
    "\n",
    "# write whole dataset containing mf and ul classes without training classes\n",
    "data_mful.to_csv(\"./Output/Daten/Hue_buildings_UTM48N_preprocessed_classes_2nd_step.csv\")\n",
    "\n",
    "# write 2nd step's training classes\n",
    "classes = gpd.read_file(training_classes_in)\n",
    "classes.FID = classes.FID.astype(np.int64)\n",
    "data_mful.FID = data_mful.FID.astype(np.int64)\n",
    "try:\n",
    "    data_mful_training_w_params = data_mful.drop('class_name', axis=1)\n",
    "except KeyError:\n",
    "    pass\n",
    "data_mful_training_w_params = data_mful_training_w_params.merge(classes[['class_name', 'FID']],\n",
    "                                                                on='FID')  # ,'class_info'\n",
    "data_mful_training_w_params.to_csv(\"./Output/Klassifikation/Hue_2nd_step_training_classes_w_params.csv\")\n",
    "\n",
    "# write new mful-reference classes with params to file\n",
    "ref = gpd.read_file(reference_classes)\n",
    "ref.FID = ref.FID.astype(np.int64)\n",
    "try:\n",
    "    data_mful_ref = ref.merge(data_mful.drop(columns='class_name', axis=1), on='FID')\n",
    "except KeyError:\n",
    "    data_mful_ref = ref.merge(data_mful, on='FID')\n",
    "data_mful_ref.to_csv('./Output/Klassifikation/Hue_reference_classes_preprocessed_mful_w_params.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "3kHDklAppb_H"
   },
   "source": [
    "<a name=\"class2\"></a>\n",
    "### Second classification\n",
    "Application of the DT-threhsold-based ruleset to classify `mf` and `ul` classes to the `data_mful` subset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "id": "m6tP7wNK9j4w"
   },
   "outputs": [],
   "source": [
    "try:\n",
    "    data_mful = data_mful.drop(columns='class_name')\n",
    "except KeyError:\n",
    "    pass\n",
    "data_mful['class_name'] = None\n",
    "\n",
    "data_mful.loc[  (data_mful.class_name.isnull() == True) &\n",
    "                (data_mful.msize_50 <= 113.07) &\n",
    "                (\n",
    "                    (data_mful.nearest == 'sf') |\n",
    "                    #(data_mful.nearest != 'sf') &\n",
    "                    (\n",
    "                        (\n",
    "                            (data_mful.tess_area <= 793.73) &\n",
    "                            (data_mful.b_class != 'sh')\n",
    "                        ) |\n",
    "                        (\n",
    "                            (data_mful.tess_area > 793.73) &\n",
    "                            # (data_mful.CAR <= 0.33) &\n",
    "                            (data_mful.perimeter <= 71.06)\n",
    "                        )\n",
    "                    )\n",
    "                ), 'class_name'] = 'mf'\n",
    "\n",
    "data_mful.loc[  (data_mful.msize_50 > 113.07) |\n",
    "                (data_mful.msize_50 <= 113.07) &\n",
    "                (data_mful.nearest != 'sf') & \n",
    "                (\n",
    "                    (\n",
    "                        (\n",
    "                            (data_mful.tess_area <= 793.73) &\n",
    "                            (data_mful.b_class == 'sh')\n",
    "                        )\n",
    "                    ) |\n",
    "                        (\n",
    "                            (data_mful.tess_area > 793.73) &\n",
    "                            (data_mful.perimeter > 71.06)  \n",
    "                        )\n",
    "                ), 'class_name'] = 'ul'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7JjWhYklfDFT"
   },
   "source": [
    "<a name=\"final\"></a>\n",
    "### Merging new information with main dataset and finalization\n",
    "The following block adds the `mf` and `ul` classes as well as the parameters that were calculated in the second step from `data_mful` to the fully classified dataset. The final result is written to the  `outfile_classified`-defined file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 66,
     "referenced_widgets": [
      "1712924346bb45bcafb9980ae20d59dc",
      "ca9974d5ca6f4eb2a2109bd1f63ea366",
      "3419e230eb524e37a769a08d8ebc49fd",
      "e60950da1bb74809912c0bc4a41006fd",
      "424192e71ebf46499d5845abcc514eaf",
      "ef82828170014116b6b6b8db84ed635d",
      "4ebcf19af3f54658bc1afcae579d4edb",
      "91bfd5d22fec4f26aac1caccde5d182b"
     ]
    },
    "executionInfo": {
     "elapsed": 299900,
     "status": "ok",
     "timestamp": 1623584666840,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "XpjidScvb_HR",
    "outputId": "b3996616-6f74-4ac4-ee48-807a8f696863"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5085/5085 [02:10<00:00, 38.99it/s]\n"
     ]
    }
   ],
   "source": [
    "# data_w_classes.FID = data_w_classes.FID.astype(np.int64)\n",
    "# data_mful.FID = data_mful.FID.astype(np.int64)\n",
    "data_classified_full = data_w_classes.copy()\n",
    "\n",
    "# data_classified_full['b_msize'] = None\n",
    "data_classified_full['nearest'] = None\n",
    "data_classified_full['mful_50m'] = None\n",
    "data_classified_full['b_class'] = None\n",
    "data_classified_full['b_mful_rat'] = None\n",
    "data_classified_full['class_50'] = None\n",
    "for i, row in tqdm(data_mful.iterrows(), total=data_mful.shape[0]):\n",
    "    # building = data_mful.loc[row.FID]\n",
    "    index = data_w_classes[data_w_classes.FID == row.FID].index\n",
    "    #     data_classified_full.loc[index, 'b_msize'] = row.b_msize\n",
    "    data_classified_full.loc[index, 'nearest'] = row.nearest\n",
    "    data_classified_full.loc[index, 'mful_50m'] = row.mful_50m\n",
    "    data_classified_full.loc[index, 'b_class'] = row.b_class\n",
    "    data_classified_full.loc[index, 'class_name'] = row.class_name\n",
    "    data_classified_full.loc[index, 'b_mful_rat'] = row.b_mful_rat\n",
    "    data_classified_full.loc[index, 'class_50'] = row.class_50\n",
    "data_classified_full.to_file(outfile_classified, encoding='utf-8')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "m5TN-phgnSrF"
   },
   "source": [
    "<a name=\"acc\"></a>\n",
    "# Accuracy assessment\n",
    "\n",
    "---\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "n_mbUN03wxVG"
   },
   "source": [
    "## Confusion matrices\n",
    "For the confusion matrix the module `sklearn.metrics` is needed. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 368
    },
    "executionInfo": {
     "elapsed": 463,
     "status": "ok",
     "timestamp": 1623584667292,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "AmFrIeQdxfFi",
    "outputId": "686cab4c-3218-45bb-bc85-47225cd44a7c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy assessment of the first step on training data\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[141,   3,   0,  11],\n",
       "       [  0,  93,   0,   2],\n",
       "       [  0,   0,  95,   0],\n",
       "       [  2,   0,   0, 103]])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# accuracy of first classification step on training data\n",
    "\n",
    "training_classes_first = gpd.read_file(training_classes_in)\n",
    "training_classes_first.FID = training_classes_first.FID.astype(np.int64)\n",
    "training_classes_first.loc[(training_classes_first.class_name == 'mf') |\n",
    "                           (training_classes_first.class_name == 'ul'),\n",
    "                           'class_name'] = 'mf_ul'\n",
    "if not 'data_classified_full' in globals():\n",
    "    data_classified_full = gpd.read_file(outfile_classified)\n",
    "    data_classified_full.FID = data_classified_full.FID.astype(np.int64)\n",
    "\n",
    "data_classified_first = data_classified_full.copy()\n",
    "data_classified_first.FID = data_classified_full.FID.astype(np.int64)\n",
    "data_classified_first.loc[(data_classified_first.class_name == 'mf') |\n",
    "                          (data_classified_first.class_name == 'ul'),\n",
    "                          'class_name'] = 'mf_ul'\n",
    "\n",
    "pred = data_classified_first[\n",
    "    data_classified_first.FID.isin(training_classes_first.FID)]\n",
    "\n",
    "# training_classes_first = training_classes_first[training_classes_first.FID.isin(pred.FID)]\n",
    "# pred = training_classes_first.drop(columns='class_name').merge(data_classified_full[['class_name', 'FID']], on='FID')\n",
    "\n",
    "labels = list(set(pred.class_name))\n",
    "confusion_matrix = metrics.confusion_matrix(training_classes_first.class_name, \n",
    "                                            pred.class_name, \n",
    "                                            normalize=None,\n",
    "                                            labels=labels)\n",
    "print(\"Accuracy assessment of the first step on training data\")\n",
    "metrics.ConfusionMatrixDisplay(confusion_matrix, display_labels=labels).plot(cmap='PuBu')\n",
    "\n",
    "confusion_matrix\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 368
    },
    "executionInfo": {
     "elapsed": 349,
     "status": "ok",
     "timestamp": 1623584667632,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "d-voJ8Bwg3bQ",
    "outputId": "afb4e993-e337-4202-eec5-3443f06249ed"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy assessment of the 2nd step on training data\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[80,  1,  6,  8],\n",
       "       [ 0,  0,  0,  0],\n",
       "       [ 7,  2, 48,  3],\n",
       "       [ 0,  0,  0,  0]])"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# accuracy of 2nd classification step on training data\n",
    "\n",
    "training_classes_mful = gpd.read_file(training_classes_in)\n",
    "training_classes_mful.FID = training_classes_mful.FID.astype(np.int64)\n",
    "training_classes_mful = training_classes_mful[(training_classes_mful.class_name == 'mf') |\n",
    "                                              (training_classes_mful.class_name == 'ul')]\n",
    "\n",
    "if not 'data_classified_full' in globals():\n",
    "    data_classified_full = gpd.read_file(outfile_classified)\n",
    "    data_classified_full.FID = data_classified_full.FID.astype(np.int64)\n",
    "\n",
    "pred = data_classified_full[data_classified_full.FID.isin(training_classes_mful.FID)]\n",
    "\n",
    "# training_classes_mful = training_classes_mful[training_classes_mful.FID.isin(pred.FID)]\n",
    "# pred = training_classes_mful.drop(columns='class_name').merge(data_classified_full[['class_name', 'FID']], on='FID')\n",
    "\n",
    "labels = list(set(pred.class_name))\n",
    "confusion_matrix = metrics.confusion_matrix(training_classes_mful.class_name, \n",
    "                                            pred.class_name, \n",
    "                                            normalize=None,\n",
    "                                            labels=labels)\n",
    "print(\"Accuracy assessment of the 2nd step on training data\")\n",
    "metrics.ConfusionMatrixDisplay(confusion_matrix, display_labels=labels).plot(cmap='PuBu')\n",
    "confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 385
    },
    "executionInfo": {
     "elapsed": 241,
     "status": "ok",
     "timestamp": 1623584667865,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "Y3bPdVL3bt6L",
    "outputId": "f33e83ca-85bf-47e9-ae0e-a96b7a66ae57"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy assessment for the fully classified dataset on reference data\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[54,  1,  0,  1,  4],\n",
       "       [ 0, 52,  0,  1,  7],\n",
       "       [ 0,  0, 60,  0,  0],\n",
       "       [ 0,  0,  0, 60,  0],\n",
       "       [ 0, 12,  0,  1, 47]])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# accuracy test on reference data\n",
    "\n",
    "ref = gpd.read_file(reference_classes)\n",
    "\n",
    "ref.FID = ref.FID.astype(np.int64)\n",
    "if not 'data_classified_full' in globals():\n",
    "    data_classified_full = gpd.read_file(outfile_classified)\n",
    "    data_classified_full.FID = data_classified_full.FID.astype(np.int64)\n",
    "\n",
    "pred = data_classified_full[data_classified_full.FID.isin(ref.FID)]\n",
    "\n",
    "# ref = ref[ref.FID.isin(data_classified_full.FID)]\n",
    "# pred = ref.drop(columns='class_name').merge(data_classified_full[['class_name', 'FID']], on='FID')\n",
    "\n",
    "labels = list(set(pred.class_name))\n",
    "confusion_matrix = metrics.confusion_matrix(ref.class_name, \n",
    "                                            pred.class_name, \n",
    "                                            normalize=None, \n",
    "                                            labels=labels)\n",
    "print(\"Accuracy assessment for the fully classified dataset on reference data\")\n",
    "metrics.ConfusionMatrixDisplay(confusion_matrix, display_labels=labels).plot(cmap='PuBu')\n",
    "confusion_matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gTaI2jAn1Gqn"
   },
   "source": [
    "<a name=\"statistics\"></a>\n",
    "# Statistics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_lVG22oGueJN"
   },
   "source": [
    "<a name=\"feature_count\"></a>\n",
    "## Feature counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 19668,
     "status": "ok",
     "timestamp": 1623584687529,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "ByyLSpV5h2mk",
    "outputId": "21391237-14b6-4cf0-e18c-3dbefe055948"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Statistics for ./Input/Hue_training_classes_no_params.csv\n",
      "us = 95 \n",
      "sh = 95 \n",
      "sf = 105 \n",
      "mf = 60 \n",
      "ul = 95 \n",
      "------------ \n",
      "sum = 108749\n",
      "\n",
      "Statistics for ./Input/Hue_reference_classes_preprocessed.csv\n",
      "us = 60 \n",
      "sh = 60 \n",
      "sf = 60 \n",
      "mf = 60 \n",
      "ul = 60 \n",
      "------------ \n",
      "sum = 108749\n",
      "\n",
      "Statistics for ./Output/Daten/Hue_buildings_UTM48N_classified_full.shp\n",
      "us = 30501 \n",
      "sh = 19343 \n",
      "sf = 53820 \n",
      "mf = 3143 \n",
      "ul = 1942 \n",
      "------------ \n",
      "sum = 108749\n",
      "\n",
      "Statistics for {'ul', 'mf'}\n",
      "us = 0 \n",
      "sh = 0 \n",
      "sf = 0 \n",
      "mf = 3143 \n",
      "ul = 1942 \n",
      "------------ \n",
      "sum = 108749\n"
     ]
    }
   ],
   "source": [
    "def statistics(filename):\n",
    "    if type(filename) == str:\n",
    "        gdf = gpd.read_file(filename)\n",
    "        print(\"\\nStatistics for\", str(filename))\n",
    "    else:\n",
    "        gdf = filename\n",
    "        print(\"\\nStatistics for\", set(data_mful.class_name))\n",
    "    us = gdf[gdf.class_name == 'us'].shape[0]\n",
    "    sh = gdf[gdf.class_name == 'sh'].shape[0]\n",
    "    sf = gdf[gdf.class_name == 'sf'].shape[0]\n",
    "    mf = gdf[gdf.class_name == 'mf'].shape[0]\n",
    "    ul = gdf[gdf.class_name == 'ul'].shape[0]\n",
    "    len(data[data.area <= 25])\n",
    "    print('us =', us,\n",
    "          '\\nsh =', sh,\n",
    "          '\\nsf =', sf,\n",
    "          '\\nmf =', mf,\n",
    "          '\\nul =', ul,\n",
    "          '\\n------------',\n",
    "          '\\nsum =', len(data))\n",
    "\n",
    "\n",
    "statistics(training_classes_in)\n",
    "\n",
    "statistics(reference_classes)\n",
    "\n",
    "statistics(outfile_classified)\n",
    "\n",
    "statistics(data_mful)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "e_Bq641McmMV"
   },
   "source": [
    "<a name=\"wards\"></a>\n",
    "## Ward related statistics\n",
    "The file `./Output/Statistik/Hue_wards_w_stats.csv` is created here. Under use of a GADM-provided file containing the adminstrative boundaries of Vietnam, the count of each class per ward of Hue is extracted. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "6uAljY09oqoN"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reprojecting to EPSG 32648\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 27/27 [00:14<00:00,  1.88it/s]\n"
     ]
    }
   ],
   "source": [
    "# Adding wards based statistics\n",
    "# Counting building types in wards\n",
    "\n",
    "wards = reproject(gpd.read_file('./Input/gadm36_VNM_3.shp'))\n",
    "wards_hue = gpd.GeoDataFrame(wards[wards.NAME_2 == 'Huế'])\n",
    "if not 'data_classified_full' in globals():\n",
    "    data_classified_full = gpd.read_file(outfile_classified)\n",
    "    data_classified_full.FID = data_classified_full.FID.astype(np.int64)\n",
    "\n",
    "for i, row in tqdm(wards_hue.iterrows(), total=wards_hue.shape[0]):\n",
    "    for class_name in ['sh', 'sf', 'mf', 'ul', 'us']:\n",
    "        wards_hue.loc[i, 'n_'+class_name] = data_classified_full[\n",
    "            data_classified_full.class_name == class_name\n",
    "            ].centroid.sindex.query(\n",
    "                row.geometry, predicate='intersects'\n",
    "                ).shape[0]\n",
    "\n",
    "wards_hue.to_csv('./Output/Statistik/Hue_wards_w_stats.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a name=\"boxplots\"></a>\n",
    "## Boxplots and histograms\n",
    "Boxplots and histograms for all the features are being generated here and written to `./Output/Statistik/`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 96144,
     "status": "ok",
     "timestamp": 1623581094467,
     "user": {
      "displayName": "Bernhard Ebersbach",
      "photoUrl": "",
      "userId": "00083767612769780957"
     },
     "user_tz": -120
    },
    "id": "JHQjU-rJymRo",
    "outputId": "ac3b150d-e49d-40e4-b51d-33debba16851"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_area.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_corners.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_perimeter.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_rect.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_shp_ratio.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_shp_index.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_area_mabr.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_width.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_length.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_elongation.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_orient.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_orient_100.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_orient_dev.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_shrd_walls.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_proximity.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_n_in_50m.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_n_in_100m.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_n_in_200m.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_n_in_ha.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_to_roads.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_roads_100.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_roads_250.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_msize_50.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_msize_100.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_tess_area.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_CAR.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_b_msize.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_b_builtup.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_b_medsize.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/boxplot_mful_50m.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/layer.py:324: PlotnineWarning: stat_boxplot : Removed 103664 rows containing non-finite values.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/cut_boxplot_area.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/layer.py:324: PlotnineWarning: stat_boxplot : Removed 102 rows containing non-finite values.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/cut_boxplot_area_mabr.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/layer.py:324: PlotnineWarning: stat_boxplot : Removed 166 rows containing non-finite values.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/cut_boxplot_b_msize.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/layer.py:324: PlotnineWarning: stat_boxplot : Removed 14 rows containing non-finite values.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/cut_boxplot_msize_50.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/layer.py:324: PlotnineWarning: stat_boxplot : Removed 31 rows containing non-finite values.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/cut_boxplot_perimeter.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/layer.py:324: PlotnineWarning: stat_boxplot : Removed 45 rows containing non-finite values.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/cut_boxplot_tess_area.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/layer.py:324: PlotnineWarning: stat_boxplot : Removed 2717 rows containing non-finite values.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/histogram_step2_mf_nearest.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/histogram_step2_ul_nearest.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/histogram_step2_mf_class_50.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/histogram_step2_ul_class_50.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/histogram_step2_mf_b_class.svg\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:719: PlotnineWarning: Saving 14.0 x 12.0 cm image.\n",
      "/home/user/miniconda3/lib/python3.9/site-packages/plotnine/ggplot.py:722: PlotnineWarning: Filename: ./Output/Statistik/histogram_step2_ul_b_class.svg\n"
     ]
    }
   ],
   "source": [
    "if not os.path.exists('./Output/Statistik'):\n",
    "    os.mkdir('./Output/Statistik')\n",
    "\n",
    "width = 14\n",
    "height = 12\n",
    "#### Statistiks for the parameters of the first step ####\n",
    "\n",
    "ref = gpd.read_file('./Input/Hue_reference_classes_preprocessed.csv')\n",
    "ref.FID = ref.FID.astype(np.int64)\n",
    "\n",
    "if not 'data_classified_full' in globals():\n",
    "    data_classified_full = gpd.read_file(outfile_classified)\n",
    "    data_classified_full.FID = data_classified_full.FID.astype(np.int64)\n",
    "\n",
    "pred = data_classified_full# data_classified_full[data_classified_full.FID.isin(ref.FID)]\n",
    "ref_w_params = pred.copy()\n",
    "\n",
    "for colname in list(pred.columns):\n",
    "    if colname not in [ 'geometry', 'FID', 'bID', 'class_name',\n",
    "                        'nearest', 'class_50', 'b_class', \n",
    "                        'b_mful_rat', 'mful_50']: \n",
    "        (\n",
    "            ggplot(pred)\n",
    "            + geom_boxplot(\n",
    "                aes(x='class_name', \n",
    "                    y=colname),\n",
    "                colour=\"#1F3552\",\n",
    "                fill=\"#4271AE\",\n",
    "                alpha=0.7,\n",
    "                outlier_shape=\".\",\n",
    "                outlier_colour=\"steelblue\",\n",
    "                width=0.5\n",
    "                )\n",
    "            #+ scale_x_discrete(labels=list(set(pred.class_name)), name='class') \n",
    "            + ggtitle(colname)\n",
    "            + theme_bw()\n",
    "        ).save(\"./Output/Statistik/boxplot_{name}.svg\".format(name=colname), units='cm', width=width, height=height)\n",
    "\n",
    "#### Statistiks for the parameters of the first step trimmed ####\n",
    "\n",
    "if not os.path.exists('./Output/Statistik'):\n",
    "    os.mkdir('./Output/Statistik')\n",
    "\n",
    "ref = gpd.read_file('./Input/Hue_reference_classes_preprocessed.csv')\n",
    "ref.FID = ref.FID.astype(np.int64)\n",
    "\n",
    "if not 'data_classified_full' in globals():\n",
    "    data_classified_full = gpd.read_file(outfile_classified)\n",
    "    data_classified_full.FID = data_classified_full.FID.astype(np.int64)\n",
    "\n",
    "pred = data_classified_full# data_classified_full[data_classified_full.FID.isin(ref.FID)]\n",
    "ref_w_params = pred.copy()\n",
    "\n",
    "# format: [param_name, max_value]\n",
    "for i in [\n",
    "          ['area', 1000], \n",
    "          ['area_mabr', 1000],\n",
    "          ['b_msize', 500], \n",
    "          ['msize_50', 500], \n",
    "          ['perimeter', 200],\n",
    "          ['tess_area', 4000]\n",
    "         ]:\n",
    "    (\n",
    "        ggplot(pred)\n",
    "        + geom_boxplot(\n",
    "            aes(x='class_name', \n",
    "                y=i[0]),\n",
    "                colour=\"#1F3552\",\n",
    "                fill=\"#4271AE\",\n",
    "                alpha=0.7,\n",
    "                outlier_shape=\".\",\n",
    "                outlier_colour=\"steelblue\",\n",
    "                width=0.5\n",
    "                )\n",
    "        #+ scale_x_discrete(labels=list(set(pred.class_name)), name='class') \n",
    "        + ggtitle(i[0])\n",
    "        + theme_bw()\n",
    "        + ylim(0, i[1])\n",
    "        + xlab(\"{title}\\n\\nNote: Outliers above {limit}\\nhave been omitted in the figure\".format(title='class_name', limit=i[1]))\n",
    "    ).save(\"./Output/Statistik/cut_boxplot_{name}.svg\".format(name=i[0]), units='cm', width=width, height=height)\n",
    "        \n",
    "        \n",
    "#### Statistiks for the parameters of the second step ####\n",
    "\n",
    "ref_2_w_params = pred[(pred.class_name == 'mf') | (pred.class_name == 'ul')].copy() \n",
    "\n",
    "for colname in ['nearest', 'class_50', 'b_class']:\n",
    "    for mful in ['mf', 'ul']:\n",
    "        (\n",
    "            ggplot(ref_2_w_params[ref_2_w_params.class_name == mful])\n",
    "            + geom_histogram(aes(x=colname), \n",
    "                                binwidth=0.5,\n",
    "                                colour=\"#1F3552\",\n",
    "                                fill=\"#4271AE\",\n",
    "                                alpha=0.7,\n",
    "                                na_rm=True)\n",
    "            + ggtitle(mful+\": \"+colname)\n",
    "            + theme_bw()\n",
    "        ).save(\"./Output/Statistik/histogram_step2_{mful}_{name}.svg\".format(mful=mful, \n",
    "                                                                    name=colname), units='cm', width=width, height=height)\n",
    "\n",
    "#ggplot(ref_2_w_params) + geom_boxplot(aes(x='class_name', y='b_mful_rat'))\n",
    "#ggplot(ref_2_w_params) + geom_boxplot(aes(x='class_name', y='mful_50m'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a name=\"resi\"></a>\n",
    "## Residential housings\n",
    "Statistical evaluation of housings per floor area"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0m² - 14m²: 0 (0.0%)\n",
      "15m² - 24m²: 0 (0.0%)\n",
      "25m² - 36m²: 12195 (16.0%)\n",
      "37m² - 48m²: 20381 (26.7%)\n",
      "49m² - 59m²: 13274 (17.4%)\n",
      "60m² - infm²: 25827 (33.8%)\n"
     ]
    }
   ],
   "source": [
    "## Distribution of residential housholds by area in Hue\n",
    "\n",
    "residential = data_classified_full[data_classified_full.class_name.isin(['mf', 'sh', 'sf'])]\n",
    "if not 'data_classified_full' in globals():\n",
    "    data_classified_full = gpd.read_file(outfile_classified)\n",
    "    data_classified_full.FID = data_classified_full.FID.astype(np.int64)\n",
    "\n",
    "def housings(data, start, stop):\n",
    "    \n",
    "    total = data[(data.area >= start) & (data.area <= stop)].shape[0]\n",
    "    perc = total / data.shape[0]\n",
    "    print(\"{start}m² - {stop}m²: {total} ({perc}%)\".format(\n",
    "        start=start,\n",
    "        stop=stop,\n",
    "        total=total,\n",
    "        perc=np.round(perc*100, 1)))\n",
    "\n",
    "for i in [[0, 14], [15, 24], [25, 36], [37, 48], [49, 59], [60, np.inf]]:\n",
    "    housings(residential, i[0], i[1])    \n",
    "          \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a name=\"means\"></a>\n",
    "## Means and standard deviation\n",
    "\n",
    "The parameter's mean values and standard deviations for each class area calculated here"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "area sh\n",
      "  Mean: 58.405346106084885\n",
      "  Std.: 23.18682141160673\n",
      "\n",
      "area sf\n",
      "  Mean: 54.07417577127462\n",
      "  Std.: 20.795062780604844\n",
      "\n",
      "area mf\n",
      "  Mean: 185.84780893095768\n",
      "  Std.: 96.77600374398948\n",
      "\n",
      "area ul\n",
      "  Mean: 381.8490201390319\n",
      "  Std.: 370.86931409426586\n",
      "\n",
      "area us\n",
      "  Mean: 20.96882325727025\n",
      "  Std.: 6.096562853277103\n",
      "__________________________\n",
      "\n",
      "\n",
      "perimeter sh\n",
      "  Mean: 34.1842715287184\n",
      "  Std.: 7.422355122624564\n",
      "\n",
      "perimeter sf\n",
      "  Mean: 29.576261463396506\n",
      "  Std.: 5.546189091684427\n",
      "\n",
      "perimeter mf\n",
      "  Mean: 60.11464432071269\n",
      "  Std.: 16.46957650617178\n",
      "\n",
      "perimeter ul\n",
      "  Mean: 87.24896680226571\n",
      "  Std.: 45.534099969119964\n",
      "\n",
      "perimeter us\n",
      "  Mean: 19.01601088357759\n",
      "  Std.: 3.209996508264248\n",
      "__________________________\n",
      "\n",
      "\n",
      "rect sh\n",
      "  Mean: 0.9867678002378121\n",
      "  Std.: 0.04016467825613592\n",
      "\n",
      "rect sf\n",
      "  Mean: 0.9840023875882571\n",
      "  Std.: 0.05177118274784027\n",
      "\n",
      "rect mf\n",
      "  Mean: 0.91324691695832\n",
      "  Std.: 0.12693205431592927\n",
      "\n",
      "rect ul\n",
      "  Mean: 0.9248803759011329\n",
      "  Std.: 0.1377828930607808\n",
      "\n",
      "rect us\n",
      "  Mean: 0.9870653742500246\n",
      "  Std.: 0.04400700787222345\n",
      "__________________________\n",
      "\n",
      "\n",
      "shp_ratio sh\n",
      "  Mean: 1.6561535521894224\n",
      "  Std.: 0.3152545171720551\n",
      "\n",
      "shp_ratio sf\n",
      "  Mean: 1.7722337484206614\n",
      "  Std.: 0.3214443000607851\n",
      "\n",
      "shp_ratio mf\n",
      "  Mean: 2.967267785555202\n",
      "  Std.: 0.7311448688061364\n",
      "\n",
      "shp_ratio ul\n",
      "  Mean: 3.869717167868177\n",
      "  Std.: 1.5483899656362141\n",
      "\n",
      "shp_ratio us\n",
      "  Mean: 1.0788747024687715\n",
      "  Std.: 0.17983489269935887\n",
      "__________________________\n",
      "\n",
      "\n",
      "shp_index sh\n",
      "  Mean: 1.2815037781109444\n",
      "  Std.: 0.09093346730912454\n",
      "\n",
      "shp_index sf\n",
      "  Mean: 1.1526218586027497\n",
      "  Std.: 0.04166275976358558\n",
      "\n",
      "shp_index mf\n",
      "  Mean: 1.2778085427935093\n",
      "  Std.: 0.15013265491688124\n",
      "\n",
      "shp_index ul\n",
      "  Mean: 1.3386177085478888\n",
      "  Std.: 0.2286652441012588\n",
      "\n",
      "shp_index us\n",
      "  Mean: 1.1860890878987576\n",
      "  Std.: 0.0691304163275362\n",
      "__________________________\n",
      "\n",
      "\n",
      "area_mabr sh\n",
      "  Mean: 59.34273552758103\n",
      "  Std.: 23.8166236873592\n",
      "\n",
      "area_mabr sf\n",
      "  Mean: 55.28634777591973\n",
      "  Std.: 22.06959535883871\n",
      "\n",
      "area_mabr mf\n",
      "  Mean: 208.04810064269807\n",
      "  Std.: 122.50789645280285\n",
      "\n",
      "area_mabr ul\n",
      "  Mean: 451.073011122554\n",
      "  Std.: 558.4084152758556\n",
      "\n",
      "area_mabr us\n",
      "  Mean: 21.306457814497882\n",
      "  Std.: 6.387520432186933\n",
      "__________________________\n",
      "\n",
      "\n",
      "width sh\n",
      "  Mean: 4.658972280928501\n",
      "  Std.: 0.9782624183503711\n",
      "\n",
      "width sf\n",
      "  Mean: 6.3390263535860285\n",
      "  Std.: 1.3366375207798045\n",
      "\n",
      "width mf\n",
      "  Mean: 10.514923592109449\n",
      "  Std.: 3.4338055868256387\n",
      "\n",
      "width ul\n",
      "  Mean: 13.147312497425334\n",
      "  Std.: 7.516478123286099\n",
      "\n",
      "width us\n",
      "  Mean: 3.5545582610406217\n",
      "  Std.: 0.7903505417124189\n",
      "__________________________\n",
      "\n",
      "\n",
      "length sh\n",
      "  Mean: 12.47731157679781\n",
      "  Std.: 3.22434542473508\n",
      "\n",
      "length sf\n",
      "  Mean: 8.497516010962467\n",
      "  Std.: 1.7755056274827232\n",
      "\n",
      "length mf\n",
      "  Mean: 19.54965023226217\n",
      "  Std.: 6.7139075394871375\n",
      "\n",
      "length ul\n",
      "  Mean: 30.170154243048405\n",
      "  Std.: 15.918405057462628\n",
      "\n",
      "length us\n",
      "  Mean: 5.99222332546474\n",
      "  Std.: 1.4004443856550381\n",
      "__________________________\n",
      "\n",
      "\n",
      "elongation sh\n",
      "  Mean: 0.13860775991314686\n",
      "  Std.: 0.02309324564675171\n",
      "\n",
      "elongation sf\n",
      "  Mean: 0.21445187829803045\n",
      "  Std.: 0.022492210430722057\n",
      "\n",
      "elongation mf\n",
      "  Mean: 0.17777394209354122\n",
      "  Std.: 0.04702555316573383\n",
      "\n",
      "elongation ul\n",
      "  Mean: 0.15518967559217303\n",
      "  Std.: 0.04972475528035152\n",
      "\n",
      "elongation us\n",
      "  Mean: 0.18872330054752304\n",
      "  Std.: 0.036303923377454785\n",
      "__________________________\n",
      "\n",
      "\n",
      "orient sh\n",
      "  Mean: 23.28801825466577\n",
      "  Std.: 14.091619982240333\n",
      "\n",
      "orient sf\n",
      "  Mean: 22.637008387959867\n",
      "  Std.: 13.646151393850907\n",
      "\n",
      "orient mf\n",
      "  Mean: 23.00692384982501\n",
      "  Std.: 14.39370861972762\n",
      "\n",
      "orient ul\n",
      "  Mean: 26.62654349639547\n",
      "  Std.: 15.024610563032\n",
      "\n",
      "orient us\n",
      "  Mean: 23.67552741221599\n",
      "  Std.: 13.867593767356885\n",
      "__________________________\n",
      "\n",
      "\n",
      "orient_100 sh\n",
      "  Mean: 23.529763706250325\n",
      "  Std.: 11.182199913191008\n",
      "\n",
      "orient_100 sf\n",
      "  Mean: 22.499158830917874\n",
      "  Std.: 10.557057749859666\n",
      "\n",
      "orient_100 mf\n",
      "  Mean: 23.029396169265034\n",
      "  Std.: 11.1742618158465\n",
      "\n",
      "orient_100 ul\n",
      "  Mean: 26.958178537590115\n",
      "  Std.: 12.03929872850289\n",
      "\n",
      "orient_100 us\n",
      "  Mean: 23.61438848201698\n",
      "  Std.: 11.069190707738487\n",
      "__________________________\n",
      "\n",
      "\n",
      "orient_dev sh\n",
      "  Mean: 5.264106152613349\n",
      "  Std.: 5.092807538056623\n",
      "\n",
      "orient_dev sf\n",
      "  Mean: 6.1537901629505765\n",
      "  Std.: 5.584730371041748\n",
      "\n",
      "orient_dev mf\n",
      "  Mean: 5.535000098631881\n",
      "  Std.: 5.491419495523926\n",
      "\n",
      "orient_dev ul\n",
      "  Mean: 4.889488208032955\n",
      "  Std.: 6.029963838239606\n",
      "\n",
      "orient_dev us\n",
      "  Mean: 5.763711603881839\n",
      "  Std.: 5.343247741305998\n",
      "__________________________\n",
      "\n",
      "\n",
      "shrd_walls sh\n",
      "  Mean: 4.988316186734219\n",
      "  Std.: 8.253904402302291\n",
      "\n",
      "shrd_walls sf\n",
      "  Mean: 2.6680787811222593\n",
      "  Std.: 4.910996313278695\n",
      "\n",
      "shrd_walls mf\n",
      "  Mean: 7.479478205536112\n",
      "  Std.: 11.508721068170976\n",
      "\n",
      "shrd_walls ul\n",
      "  Mean: 9.02574665293512\n",
      "  Std.: 14.58415531792853\n",
      "\n",
      "shrd_walls us\n",
      "  Mean: 1.9747549260679977\n",
      "  Std.: 3.5729392565626688\n",
      "__________________________\n",
      "\n",
      "\n",
      "n_in_50m sh\n",
      "  Mean: 52.46668045287701\n",
      "  Std.: 26.813976070321644\n",
      "\n",
      "n_in_50m sf\n",
      "  Mean: 46.94318097361575\n",
      "  Std.: 27.515635313137306\n",
      "\n",
      "n_in_50m mf\n",
      "  Mean: 51.01813553929367\n",
      "  Std.: 25.26271737257996\n",
      "\n",
      "n_in_50m ul\n",
      "  Mean: 30.183831101956745\n",
      "  Std.: 22.487797037520632\n",
      "\n",
      "n_in_50m us\n",
      "  Mean: 49.28841677322055\n",
      "  Std.: 27.650006536497322\n",
      "__________________________\n",
      "\n",
      "\n",
      "n_in_100m sh\n",
      "  Mean: 152.73861345189474\n",
      "  Std.: 77.34936712350597\n",
      "\n",
      "n_in_100m sf\n",
      "  Mean: 137.83963210702342\n",
      "  Std.: 80.14005448387051\n",
      "\n",
      "n_in_100m mf\n",
      "  Mean: 150.21762647152403\n",
      "  Std.: 73.43318780651224\n",
      "\n",
      "n_in_100m ul\n",
      "  Mean: 94.12100926879506\n",
      "  Std.: 65.55289121199942\n",
      "\n",
      "n_in_100m us\n",
      "  Mean: 144.56857152224518\n",
      "  Std.: 80.1953611627459\n",
      "__________________________\n",
      "\n",
      "\n",
      "n_in_200m sh\n",
      "  Mean: 484.94457943442075\n",
      "  Std.: 241.8072737061111\n",
      "\n",
      "n_in_200m sf\n",
      "  Mean: 435.1422891118543\n",
      "  Std.: 249.20952802457373\n",
      "\n",
      "n_in_200m mf\n",
      "  Mean: 488.14285714285717\n",
      "  Std.: 224.30509571725935\n",
      "\n",
      "n_in_200m ul\n",
      "  Mean: 351.0834191555098\n",
      "  Std.: 213.62730726409183\n",
      "\n",
      "n_in_200m us\n",
      "  Mean: 452.8761024228714\n",
      "  Std.: 245.04904131176818\n",
      "__________________________\n",
      "\n",
      "\n",
      "n_in_ha sh\n",
      "  Mean: 37.71850281755674\n",
      "  Std.: 19.322611021502507\n",
      "\n",
      "n_in_ha sf\n",
      "  Mean: 33.63175399479747\n",
      "  Std.: 19.71448315358686\n",
      "\n",
      "n_in_ha mf\n",
      "  Mean: 36.76932866687878\n",
      "  Std.: 18.336860669310084\n",
      "\n",
      "n_in_ha ul\n",
      "  Mean: 21.83522142121524\n",
      "  Std.: 16.376700796175353\n",
      "\n",
      "n_in_ha us\n",
      "  Mean: 35.304317891216684\n",
      "  Std.: 19.768954390598775\n",
      "__________________________\n",
      "\n",
      "\n",
      "to_roads sh\n",
      "  Mean: 24.156605811404646\n",
      "  Std.: 45.535810195702275\n",
      "\n",
      "to_roads sf\n",
      "  Mean: 31.597728620958748\n",
      "  Std.: 58.35577298209173\n",
      "\n",
      "to_roads mf\n",
      "  Mean: 20.422315482023546\n",
      "  Std.: 30.624870693064004\n",
      "\n",
      "to_roads ul\n",
      "  Mean: 31.87604443357363\n",
      "  Std.: 50.993853145469124\n",
      "\n",
      "to_roads us\n",
      "  Mean: 28.535404653617917\n",
      "  Std.: 56.6683943484577\n",
      "__________________________\n",
      "\n",
      "\n",
      "roads_100 sh\n",
      "  Mean: 10.793982319185234\n",
      "  Std.: 9.276216028933064\n",
      "\n",
      "roads_100 sf\n",
      "  Mean: 9.147417316982535\n",
      "  Std.: 8.902348542326608\n",
      "\n",
      "roads_100 mf\n",
      "  Mean: 11.623608017817372\n",
      "  Std.: 9.429169414784809\n",
      "\n",
      "roads_100 ul\n",
      "  Mean: 10.110710607621009\n",
      "  Std.: 8.90229599341412\n",
      "\n",
      "roads_100 us\n",
      "  Mean: 9.840005245729648\n",
      "  Std.: 9.366797812955253\n",
      "__________________________\n",
      "\n",
      "\n",
      "roads_250 sh\n",
      "  Mean: 46.90280721708111\n",
      "  Std.: 37.09301425382936\n",
      "\n",
      "roads_250 sf\n",
      "  Mean: 39.453790412486065\n",
      "  Std.: 36.08188334056823\n",
      "\n",
      "roads_250 mf\n",
      "  Mean: 49.63633471205854\n",
      "  Std.: 36.155888534458086\n",
      "\n",
      "roads_250 ul\n",
      "  Mean: 44.13233779608651\n",
      "  Std.: 31.962186312187686\n",
      "\n",
      "roads_250 us\n",
      "  Mean: 43.295367365004424\n",
      "  Std.: 38.09808504775164\n",
      "__________________________\n",
      "\n",
      "\n",
      "msize_50 sh\n",
      "  Mean: 52.480655895672854\n",
      "  Std.: 17.45797148861496\n",
      "\n",
      "msize_50 sf\n",
      "  Mean: 49.16062752601263\n",
      "  Std.: 16.763598463623932\n",
      "\n",
      "msize_50 mf\n",
      "  Mean: 69.88591804963411\n",
      "  Std.: 18.752638825420686\n",
      "\n",
      "msize_50 ul\n",
      "  Mean: 160.48331575180225\n",
      "  Std.: 165.45254855943492\n",
      "\n",
      "msize_50 us\n",
      "  Mean: 42.97796742631389\n",
      "  Std.: 14.785621244036482\n",
      "__________________________\n",
      "\n",
      "\n",
      "msize_100 sh\n",
      "  Mean: 53.956138109393585\n",
      "  Std.: 16.300674596713876\n",
      "\n",
      "msize_100 sf\n",
      "  Mean: 51.100016185990334\n",
      "  Std.: 18.139002723651277\n",
      "\n",
      "msize_100 mf\n",
      "  Mean: 71.10515501431753\n",
      "  Std.: 27.038599199261316\n",
      "\n",
      "msize_100 ul\n",
      "  Mean: 133.17633814624097\n",
      "  Std.: 82.9480708395178\n",
      "\n",
      "msize_100 us\n",
      "  Mean: 46.44831319137078\n",
      "  Std.: 16.317631708883685\n",
      "__________________________\n",
      "\n",
      "\n",
      "tess_area sh\n",
      "  Mean: 641.3376802491858\n",
      "  Std.: 4826.665361765669\n",
      "\n",
      "tess_area sf\n",
      "  Mean: 861.4818108097361\n",
      "  Std.: 6985.443286890104\n",
      "\n",
      "tess_area mf\n",
      "  Mean: 921.4373322367165\n",
      "  Std.: 3138.2324777915883\n",
      "\n",
      "tess_area ul\n",
      "  Mean: 3166.275147106076\n",
      "  Std.: 30068.49837847568\n",
      "\n",
      "tess_area us\n",
      "  Mean: 968.2104793626438\n",
      "  Std.: 11524.690267095622\n",
      "__________________________\n",
      "\n",
      "\n",
      "CAR sh\n",
      "  Mean: 0.3241631871995037\n",
      "  Std.: 0.20612021328249547\n",
      "\n",
      "CAR sf\n",
      "  Mean: 0.2788095185804534\n",
      "  Std.: 0.19549692049120024\n",
      "\n",
      "CAR mf\n",
      "  Mean: 0.39026393891186767\n",
      "  Std.: 0.20018383297273967\n",
      "\n",
      "CAR ul\n",
      "  Mean: 0.3378318280123584\n",
      "  Std.: 0.19839730566041291\n",
      "\n",
      "CAR us\n",
      "  Mean: 0.24034265925707354\n",
      "  Std.: 0.195448838553181\n",
      "__________________________\n",
      "\n",
      "\n",
      "bID sh\n",
      "  Mean: 585.9932275241689\n",
      "  Std.: 330.8202111468274\n",
      "\n",
      "bID sf\n",
      "  Mean: 612.9762170196952\n",
      "  Std.: 329.94764444983804\n",
      "\n",
      "bID mf\n",
      "  Mean: 560.9688195991091\n",
      "  Std.: 334.20484598325226\n",
      "\n",
      "bID ul\n",
      "  Mean: 574.6127703398558\n",
      "  Std.: 321.59905954392076\n",
      "\n",
      "bID us\n",
      "  Mean: 623.0673092685486\n",
      "  Std.: 332.205134165619\n",
      "__________________________\n",
      "\n",
      "\n",
      "b_msize sh\n",
      "  Mean: 55.66271200279171\n",
      "  Std.: 14.947093595757906\n",
      "\n",
      "b_msize sf\n",
      "  Mean: 53.72231497491638\n",
      "  Std.: 18.803533360667522\n",
      "\n",
      "b_msize mf\n",
      "  Mean: 80.53922508113268\n",
      "  Std.: 58.98534533371427\n",
      "\n",
      "b_msize ul\n",
      "  Mean: 137.6325962821833\n",
      "  Std.: 166.43385320629696\n",
      "\n",
      "b_msize us\n",
      "  Mean: 49.729882102226156\n",
      "  Std.: 18.771212079032455\n",
      "__________________________\n",
      "\n",
      "\n",
      "b_builtup sh\n",
      "  Mean: 1009.1063471167863\n",
      "  Std.: 1404.635121345031\n",
      "\n",
      "b_builtup sf\n",
      "  Mean: 1215.4896297264959\n",
      "  Std.: 1608.0075385300895\n",
      "\n",
      "b_builtup mf\n",
      "  Mean: 937.3425393923003\n",
      "  Std.: 1327.200378817325\n",
      "\n",
      "b_builtup ul\n",
      "  Mean: 913.0681722863028\n",
      "  Std.: 1209.7414305614648\n",
      "\n",
      "b_builtup us\n",
      "  Mean: 972.074632116652\n",
      "  Std.: 1376.8007180038678\n",
      "__________________________\n",
      "\n",
      "\n",
      "b_medsize sh\n",
      "  Mean: 44.70641996484516\n",
      "  Std.: 9.901905531031206\n",
      "\n",
      "b_medsize sf\n",
      "  Mean: 43.47137332218507\n",
      "  Std.: 10.540601617760307\n",
      "\n",
      "b_medsize mf\n",
      "  Mean: 54.76169617562837\n",
      "  Std.: 23.387669184650605\n",
      "\n",
      "b_medsize ul\n",
      "  Mean: 88.7204033161689\n",
      "  Std.: 151.5674015745069\n",
      "\n",
      "b_medsize us\n",
      "  Mean: 39.76648849873774\n",
      "  Std.: 10.04232876089025\n",
      "__________________________\n",
      "\n"
     ]
    }
   ],
   "source": [
    "## Calculation of parameters' mean values and standard deviation per class\n",
    "if not 'data_classified_full' in globals():\n",
    "    data_classified_full = gpd.read_file(outfile_classified)\n",
    "    data_classified_full.FID = data_classified_full.FID.astype(np.int64)\n",
    "\n",
    "for i in data_classified_full.columns:\n",
    "    if type(data_classified_full[data_classified_full.class_name == j][i].values[1]) == np.float64:\n",
    "        for j in set(data_classified_full.class_name):\n",
    "            print('\\n'+i, j)\n",
    "            print(\"  Mean:\", np.mean(data_classified_full[data_classified_full.class_name == j][i].values))\n",
    "            print(\"  Std.:\", np.std(data_classified_full[data_classified_full.class_name == j][i].values))\n",
    "        else:\n",
    "            pass\n",
    "        print('__________________________\\n')   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [
    "GVN1aUjz546e",
    "mrL0P2-BcPYm"
   ],
   "name": "Hue_classification.ipynb",
   "provenance": [
    {
     "file_id": "1kR-f6rJ5XhL-VUsFGiQP7wiUJqdUTyU4",
     "timestamp": 1623590366189
    }
   ]
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.5"
  },
  "toc-autonumbering": true,
  "toc-showcode": false,
  "toc-showmarkdowntxt": false,
  "toc-showtags": false,
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "02248b97e95e471896dec3c2c2445c2c": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "046afab5c04643cfba91480444490605": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "054ccda05e8248bc97004e0950053b3a": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "166973b412084481849fc6b37c431494": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "16d1d3418e3a47e4aef4ff040c1f0a64": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_b6bcc204e49e486ba2567c556069bd91",
      "max": 5085,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_166973b412084481849fc6b37c431494",
      "value": 5085
     }
    },
    "1712924346bb45bcafb9980ae20d59dc": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_3419e230eb524e37a769a08d8ebc49fd",
       "IPY_MODEL_e60950da1bb74809912c0bc4a41006fd"
      ],
      "layout": "IPY_MODEL_ca9974d5ca6f4eb2a2109bd1f63ea366"
     }
    },
    "1c2af11f863a4e89b769f13c24ccdcbc": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "201a57ace6c14df68f078c6db7411c2e": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_d2ff2ee31b1b4f19bccb592657fbafba",
       "IPY_MODEL_7c509b8e62224a36af1311f3838a82bf"
      ],
      "layout": "IPY_MODEL_d524c1c6f0aa45a787571ddfde83f68a"
     }
    },
    "246d31140fd74d7ab95d22fc2b276987": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_552afccb92824ee99784496a51c89561",
       "IPY_MODEL_356f0e34ef7d446992147c1450868b0c"
      ],
      "layout": "IPY_MODEL_6a9f009e7a9f4019a51dd6ad0adf9b9d"
     }
    },
    "2c7f4be35c9143459746b4d16a529381": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "3280f822d845445abb1f49788ce46b5a": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "3419e230eb524e37a769a08d8ebc49fd": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_ef82828170014116b6b6b8db84ed635d",
      "max": 5085,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_424192e71ebf46499d5845abcc514eaf",
      "value": 5085
     }
    },
    "356f0e34ef7d446992147c1450868b0c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_bb0c7671947e424e884441b31983fd2b",
      "placeholder": "​",
      "style": "IPY_MODEL_eae35f5a58894102b5e0d4d4efbf8d66",
      "value": " 1102/1102 [01:01&lt;00:00, 17.94it/s]"
     }
    },
    "360d95d8a9114dc3a2117da84a5f81a2": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "424192e71ebf46499d5845abcc514eaf": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "47053e3b47d74f5ea758c34e3a07a76b": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "4ebcf19af3f54658bc1afcae579d4edb": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "512cc211d14548b5b90adce91b7a0f5e": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "552afccb92824ee99784496a51c89561": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_02248b97e95e471896dec3c2c2445c2c",
      "max": 1102,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_512cc211d14548b5b90adce91b7a0f5e",
      "value": 1102
     }
    },
    "59442c7e7a4f4ef4928e1b3f7cd71843": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "624d5c75304d4db3923b438a33774434": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "65b11cea931f425eb0ee046ed35ccef0": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_3280f822d845445abb1f49788ce46b5a",
      "max": 5085,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_624d5c75304d4db3923b438a33774434",
      "value": 5085
     }
    },
    "6a9f009e7a9f4019a51dd6ad0adf9b9d": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "6ac3313a05884052b86ce3dd001ff160": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_47053e3b47d74f5ea758c34e3a07a76b",
      "placeholder": "​",
      "style": "IPY_MODEL_360d95d8a9114dc3a2117da84a5f81a2",
      "value": " 5085/5085 [03:02&lt;00:00, 27.84it/s]"
     }
    },
    "7c509b8e62224a36af1311f3838a82bf": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_1c2af11f863a4e89b769f13c24ccdcbc",
      "placeholder": "​",
      "style": "IPY_MODEL_8f1f870dad824d059540b879488a4ae5",
      "value": " 5085/5085 [04:40&lt;00:00, 18.14it/s]"
     }
    },
    "8f1f870dad824d059540b879488a4ae5": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "91bfd5d22fec4f26aac1caccde5d182b": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "aae6bd861fb54b3ebef1e5acdde347c1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "b6bcc204e49e486ba2567c556069bd91": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "bb0c7671947e424e884441b31983fd2b": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "c7f2dc2a44e6440cb282be20c57d3017": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_2c7f4be35c9143459746b4d16a529381",
      "placeholder": "​",
      "style": "IPY_MODEL_aae6bd861fb54b3ebef1e5acdde347c1",
      "value": " 5085/5085 [00:07&lt;00:00, 671.67it/s]"
     }
    },
    "ca9974d5ca6f4eb2a2109bd1f63ea366": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d0802212b19043c3be7b4d9235dcc495": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d2ff2ee31b1b4f19bccb592657fbafba": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_d0802212b19043c3be7b4d9235dcc495",
      "max": 5085,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_046afab5c04643cfba91480444490605",
      "value": 5085
     }
    },
    "d524c1c6f0aa45a787571ddfde83f68a": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "e14e97bc5c15410385ffcc93c67ca4f1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_65b11cea931f425eb0ee046ed35ccef0",
       "IPY_MODEL_c7f2dc2a44e6440cb282be20c57d3017"
      ],
      "layout": "IPY_MODEL_054ccda05e8248bc97004e0950053b3a"
     }
    },
    "e60950da1bb74809912c0bc4a41006fd": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_91bfd5d22fec4f26aac1caccde5d182b",
      "placeholder": "​",
      "style": "IPY_MODEL_4ebcf19af3f54658bc1afcae579d4edb",
      "value": " 5085/5085 [02:53&lt;00:00, 29.27it/s]"
     }
    },
    "eae35f5a58894102b5e0d4d4efbf8d66": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "ef82828170014116b6b6b8db84ed635d": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "f58106c95326412db9edf3eefbc89140": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_16d1d3418e3a47e4aef4ff040c1f0a64",
       "IPY_MODEL_6ac3313a05884052b86ce3dd001ff160"
      ],
      "layout": "IPY_MODEL_59442c7e7a4f4ef4928e1b3f7cd71843"
     }
    }
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}