Predict Customer Churn

• Project Predict Customer Churn of ML DevOps Engineer Nanodegree Udacity

Project Description

Python library which provides functions to analyze, feature engineer and train models to predict bank customer churn.

Files and data description

Overview of the files and data present in the root directory.

- data			# folder for input data
  - bank_data.csv
- images
  - eda 		# images from eda phase
  - results		# report images from training process
- logs 			# log file output from running the training
- models		# output directory for trained models
- churn_library.py	# python library
- churn_script_logging_and_tests.py # executable to run tests and train models
- churn_notebook.ipynb	# jupyter notebook to fiddle with data

Install dependencies

Install module dependencies

   python3 -m pip install -r requirements_py3.8.txt

You should also install autopep8

   python3 -m pip install autopep8

and pylint

   python3 -m pip install pylint

Running Files

Below command

   ipython churn_script_logging_and_tests.py

will run tests on the library and if successful perform the training process using the functions provided in churn_library.py. If there are any errors the command will fail. Details of the error can be found in the ./logs directory. If the command runs successfully it will produce trained models in the ./models directory and quality metrics of the models in ./images/results