-
Notifications
You must be signed in to change notification settings - Fork 0
/
article_ml.py
238 lines (185 loc) · 7.95 KB
/
article_ml.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
import matplotlib.pyplot as plt
from sklearn.feature_selection import SelectKBest
from sklearn.feature_selection import f_regression
from sklearn.feature_selection import RFE
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_absolute_error
import pandas as pd
import numpy as np
def plot_squareMeters(df):
plt.scatter(df[['squareMeters']], df[['price']], alpha = 0.5)
plt.xlabel('squareMeters')
plt.ylabel('price')
plt.show()
def plot_features(df):
df.hist(bins = 50, figsize = (20,20))
plt.show()
def print_df_info(df):
print("\nDF.head():")
print(df.head())
print("\nDF.info():")
print(df.info())
print("\nDF.describe():")
print(df.describe())
def get_Xy(df):
X = df.reindex(columns = df.columns[:-1]); X = X.values.tolist();
y = df.reindex(columns = [df.columns[-1]]); y = y.values.tolist();
return X,y
def get_dict_features(X, feature_names):
dict_features = {}
dict_features['squareMeters'] = [[Xi[0] for Xi in X]]
dict_features['numberOfRooms'] = [[Xi[1] for Xi in X]]
dict_features['hasYard'] = [[Xi[2] for Xi in X]]
dict_features['hasPool'] = [[Xi[3] for Xi in X]]
dict_features['floors'] = [[Xi[4] for Xi in X]]
dict_features['cityCode'] = [[Xi[5] for Xi in X]]
dict_features['cityPartRange'] = [[Xi[6] for Xi in X]]
dict_features['numPrevOwners'] = [[Xi[7] for Xi in X]]
dict_features['made'] = [[Xi[8] for Xi in X]]
dict_features['isNewBuilt'] = [[Xi[9] for Xi in X]]
dict_features['hasStormProtector'] = [[Xi[10] for Xi in X]]
dict_features['basement'] = [[Xi[11] for Xi in X]]
dict_features['attic'] = [[Xi[12] for Xi in X]]
dict_features['garage'] = [[Xi[13] for Xi in X]]
dict_features['hasStorageRoom'] = [[Xi[14] for Xi in X]]
dict_features['hasGuestRoom'] = [[Xi[15] for Xi in X]]
return dict_features
def get_regression(X,y):
X_train, X_test, y_train, y_test = train_test_split(X,y,test_size=0.3, random_state=1)
reg = LinearRegression()
reg.fit(X_train, y_train)
return reg, X_train, X_test, y_train, y_test
def prediction(reg, X_test, y_test):
y_pred = reg.predict(X_test)
mse = mean_squared_error(y_test, y_pred, squared=False)
mae = mean_absolute_error(y_test, y_pred)
return mae
def print_res(y_test, y_pred):
intercept, slope, corr_coeff = plot_linear_regression(y_test, y_pred)
plt.show()
""" Lance le test l'élimination de features récursif """
def test_rfe(train_set, label_train, lin_reg):
res = []
for i in range(1,17):
rfe_i = RFE(lin_reg, n_features_to_select=i, step=1)
rfe_i = rfe_i.fit(train_set, label_train)
predictions_rfe_i = rfe_i.predict(train_set)
lin_mse_rfe_i = mean_squared_error(label_train, predictions_rfe_i)
lin_rmse_rfe_i = np.sqrt(lin_mse_rfe_i)
print(i," ",lin_rmse_rfe_i)
res.append(lin_rmse_rfe_i)
plt.plot(range(1,17),res)
plt.ylim([0,7000])
plt.show()
""" Renvoie les meilleurs features choisies par SelectKBest() """
def get_best_features(X_train, y_train, X_test, feature_names, n_feature):
fs = SelectKBest(score_func=f_regression, k='all')
fs.fit(X_train, y_train)
for i in range(len(fs.scores_)):
print(f'{feature_names[i]}: {fs.scores_[i]}')
# plot the scores
plt.bar([i for i in range(len(fs.scores_))], fs.scores_)
plt.ylim([0,10])
plt.show()
coef_copy = fs.scores_.copy()
coef_copy.sort()
best_f = []
coef_copy = coef_copy[::-1]
for i in range(n_feature):
best_f.append(np.where(fs.scores_==coef_copy[i]))
best_features = [feature_names[i[0][0]] for i in best_f]
return best_features
""" Filtre les features """
def filter_features(feature_names, best_features, df):
not_chosen_features = [x for x in feature_names if not x in best_features]
print("features_names again and again: ",feature_names)
print(f"\n\tSupprimons les features:\n{not_chosen_features}")
for feature in not_chosen_features:
df.pop(feature)
feature_names, _ = get_feature_target_names(df)
X, y = get_Xy(df)
print(f"dimension de X: {len(X[0])}")
return X,y,feature_names
""" Lance le test ou la moins bonne des features est retirée jusqu'à ce qu'il n'en reste qu'une """
def plot_with_best_features(features_start, df, n_feature_min=1):
nb_feature = len(features_start)
results = []
X,y = get_Xy(df)
print(f"features: {features_start}\nlen: {nb_feature}")
print(range(nb_feature))
for i in range(nb_feature):
reg, X_train, X_test, y_train, y_test = get_regression(X,y)
best_features = get_best_features(X_train, y_train, X_test, features_start, len(features_start)-1)
X,y, features_start = filter_features(features_start, best_features, df)
result = prediction(reg,X_test, y_test)
results.append(result)
print(f"\nFor {len(features_start)} features, precision is {result}")
if len(X[0]) < n_feature_min:
break
list_features_descending = list(reversed(range(1,nb_feature+1)))
print(f"x:{list_features_descending}\nresults: {results}")
plt.plot(list_features_descending, results)
plt.show()
""" Renvoie le nom des features et le nom de la cible en assumant que
la dernière colonne est la cible """
def get_feature_target_names(dataframe, wanted_features = None):
# split the last element of the dataframe. (-1 -> len()-1)
columns = dataframe.columns
if wanted_features == None:
feature_names = list(columns[:-1])
else:
feature_names = list(set(columns).intersection(set(wanted_features)))
target_name = columns[-1]
return feature_names, target_name
""" Présente les différentes fonctionalités et plus montrées dans l'article """
def test_article():
""" On récupère le jeu de données Xy depuis le fichier .csv """
df = pd.read_csv("ParisHousing.csv")
feature_names, target_name = get_feature_target_names(df)
X, y = get_Xy(df)
X = np.array(X)
y = np.array(y)
print("Voici nos features")
plot_features(df)
print("Voici les informations concernant notre base de donnée")
print_df_info(df)
print("Courbe de y = x * prix du mètre au carré")
plot_squareMeters(df)
""" Split de la data + entraînement du modèle """
X_train, X_test, y_train, y_test = train_test_split(X,y, test_size = 0.3)
reg = LinearRegression()
reg.fit(X_train, y_train)
y_pred = reg.predict(X_test)
mae = mean_absolute_error(y_test, y_pred)
print(f"mean absolute error : {mae}")
"""
Calcul des meilleurs features + affichage matplotlib """
fs = SelectKBest(score_func=f_regression, k='all')
fs.fit(X_train, y_train)
for i in range(len(fs.scores_)):
print(f'{feature_names[i]}: {fs.scores_[i]}')
# plot the scores
plt.bar([i for i in range(len(fs.scores_))], fs.scores_)
plt.ylim([0,10])
plt.xlabel("features")
plt.ylabel("coef")
plt.show()
""" Elimination des features moins efficaces et affichage resultat """
# Prenons les 10 meilleurs features:
to_remove = []
for i in range(len(fs.scores_)):
if fs.scores_[i] < 0.30:
to_remove.append(feature_names[i])
filtered_features = [feature for feature in feature_names if not feature in to_remove]
X = df.reindex(columns =filtered_features)
X = X.values.tolist()
print(f"nouvelle dimension de X: {len(X[0])}")
print(f"Features choisies:\n{filtered_features}")
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3)
y_pred = LinearRegression().fit(X_train, y_train).predict(X_test)
mae = mean_absolute_error(y_test, y_pred)
print(f"resultat avec réduction de features: {mae}")
""" L'interpreteur commencera ici """
if __name__ == "__main__":
test_article()