-
Notifications
You must be signed in to change notification settings - Fork 0
/
Official Code
311 lines (257 loc) · 10.9 KB
/
Official Code
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
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from pandas import ExcelWriter
import plotly.graph_objs as go
import plotly.offline as offline
import plotly.figure_factory as ff
import seaborn as sns
import mglearn
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from imblearn.over_sampling import SMOTE
from sklearn.model_selection import GridSearchCV
from sklearn.ensemble import RandomForestClassifier
from sklearn import tree
from sklearn import metrics
from pprint import pprint
from sklearn.metrics import classification_report
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.model_selection import cross_val_score
from sklearn.model_selection import cross_val_predict
from sklearn.metrics import confusion_matrix
path = r'C:\\Users\\giovanni\\Desktop\\FP_mnist\\hr_churn\\hr_emp_attrition_original.xlsx'
#Store dataset into df
df = pd.read_excel(path)
################### Data Exploration ###################
print(df.shape)
print(df.info())
print(df.describe())
dtype_summary = df.dtypes.value_counts()
dtype_summary = pd.Series(dtype_summary)
#Writing data set description into excel
with pd.ExcelWriter (r'C:\\Users\\giovanni\\Desktop\\FP_mnist\\hr_churn\\hr_employee_attrition_description.xlsx') as writer:
df.describe().to_excel(writer,sheet_name = 'description')
dtype_summary.to_excel(writer,sheet_name = 'dtype_distribution',header = False)
print("Distribution of Data Types: \n",dtype_summary)
#### Challenge #### = #I want to plot a table ...
#Check for NA
check_for_na = df.isnull().values.any()
print("Null values within dataset: ",check_for_na)
#Calculating % of each class
label_distribution = pd.Series(df['Attrition']).value_counts()
no_class = label_distribution[0]
yes_class =label_distribution[1]
total_entries = label_distribution[0]+label_distribution[1]
no_percent = label_distribution[0]/total_entries
yes_percent = label_distribution[1]/total_entries
print("{0:.0f}%".format(no_percent*100))
print("{0:.0f}%".format(yes_percent*100))
######################################################################################## Visualisations #######################################################################################
## The goal of data viz is to better understand what is behind the data (scenario description)
#Bar Plot Attrition
x=['No','Yes']
y = [no_class,yes_class]
fig = plt.figure(figsize = (9,6))
ax = plt.subplot(111)
ax.bar(x, y, color =('#2680eb','#e34850'))
for index,value in enumerate(y):
ax.annotate(value, xy=(index,value))
plt.show()
#Pie Chart Distribution of Attrition
labels = 'No', 'Yes'
x_percent = [no_class,yes_class]
fig_pie_chart = plt.figure(figsize = (9,6))
fig_pie_chart, ax1 = plt.subplots()
ax1.pie(x_percent, labels = labels,colors =('#2680eb','#e34850') ,autopct = '%1.0f%%',startangle = 90)
ax1.axis('equal')
plt.show()
#Gender distribution
gender_count = pd.crosstab(df['Gender'], df['Attrition'])
gender_count.plot.bar(stacked = True,color =('#2680eb','#e34850'))
plt.show()
#Job Role Distribution
job_role_count = pd.crosstab(df['JobRole'], df['Attrition'])
job_role_count.plot.bar(stacked = True,color =('#2680eb','#e34850'))
plt.show()
#Job Role and Gender
cross_tab = pd.crosstab(df['JobRole'], df['Gender'])
cross_tab.plot.bar(stacked = True, color =('#FFC0CB','#2680eb'))
plt.show()
#Statistics about job satisfaction
job_satisfaction_avg = df['JobSatisfaction'].median()
job_satisfaction_avg_box_plot = sns.boxplot(y= df['JobSatisfaction'])
plt.show()
########### Statistics about monthly salary ###########
#Monthly salary by Gender
palette = {"Male": "#2680eb", "Female":"#FFC0CB"}
sns.boxplot(y = df['Gender'],x = df['MonthlyIncome'],palette = palette)
plt.show()
#Box Plot of Monthly Income group by gender
ax = sns.boxplot(y = df['MonthlyIncome'],x = df['Gender'],palette = palette)
medians = df.groupby(['Gender'])['MonthlyIncome'].median().values
median_labels = [str(np.round(s, 2)) for s in medians]
print(median_labels)
pos = range(len(medians))
for tick, label in zip(pos,ax.get_xticklabels()):
ax.text(pos[tick],medians[tick]+0.03,median_labels[tick],
horizontalalignment='center', size='x-large', color='w', weight='bold')
plt.show()
#Salary By Job Roles
ax = sns.boxplot(x = df['MonthlyIncome'],y = df['JobRole'])
medians = df.groupby(['JobRole'])['MonthlyIncome'].median().values
median_labels = [str(np.round(s, 2)) for s in medians]
print(median_labels)
pos = range(len(medians))
for tick, label in zip(pos,ax.get_xticklabels()):
ax.text(pos[tick],medians[tick]+0.03,median_labels[tick],
verticalalignment='top', size='smaller', color='b', weight='bold')
plt.show()
##################################################
### To do:
#1: Add text to each value and column
#2: Attrition rate by jobsatisfaction
#3: Plot age distribution
#3: Sort columns
#4: What is the gender/job role with the highest attrition rate?
#################################################
########Correlation HeatMap########
#Plotting Correlation Heatmap
#Converting categorical into numerical variables
def converter (col):
if col == 'Yes':
return 1
else:
return 0
#dropping variables which are not contributing to the analysis
df['Attrition'] = df['Attrition'].apply(converter)
df['OverTime'] = df['OverTime'].apply(converter)
categorical_features = ['BusinessTravel','Department','EducationField','Gender','JobRole','MaritalStatus']
# Correlation Heatmap for feature selection
corrs = df.corr()
figure = ff.create_annotated_heatmap(
z = corrs.values,
x = list(corrs.columns),
y = list(corrs.index),
annotation_text = corrs.round(2).values,
showscale = True)
offline.plot(figure, filename = 'corrheatmap.html')
######################################################################################## Random Forest with Cross Validation #######################################################################################
df = df.drop(['Over18','EmployeeNumber','StandardHours','EmployeeCount','MonthlyRate','NumCompaniesWorked','PerformanceRating','Age','DistanceFromHome','YearsAtCompany'],axis = 1)
categorical_features = ['BusinessTravel','Department','EducationField','Gender','JobRole','MaritalStatus']
df_ml = pd.get_dummies(df, columns = categorical_features)
#Dividing the data in label and features
X = df_ml.drop(['Attrition'],axis = 1 )#Features
Y = df_ml['Attrition']#Label
#Normalise dataset
feature_scaler = StandardScaler()
X_scaled = feature_scaler.fit_transform(X)
#I want to up-sample attrition = 'Yes' stored in Y. I will perform the process only on the train set
print("Number of observations in each class before oversampling (training data): \n", pd.Series(Y).value_counts())
smote = SMOTE(random_state = 101)
X_scaled, Y = smote.fit_sample(X_scaled, Y)
print("Number of observations in each class after oversampling (training data): \n", pd.Series(Y).value_counts())
#Split into train and test
X_train, X_test, Y_train, Y_test = train_test_split( X_scaled, Y, test_size = 0.3, random_state = 100)
#Implementing random forest
rfc = RandomForestClassifier(criterion = 'entropy')
rfc.fit(X_train,Y_train)
Y_pred = rfc.predict(X_test)
print("Prediction Accuracy: ", metrics.recall_score(Y_test, Y_pred))
conf_mat = metrics.confusion_matrix(Y_test, Y_pred)
plt.figure(figsize=(8,6))
sns.heatmap(conf_mat,annot=True)
plt.title("Confusion_matrix")
plt.xlabel("Predicted Class")
plt.ylabel("Actual class")
plt.show()
print('Confusion matrix: \n', conf_mat)
print('TP: ', conf_mat[1,1])
print('TN: ', conf_mat[0,0])
print('FP: ', conf_mat[0,1])
print('FN: ', conf_mat[1,0])
#Grid search implementation
param_grid = {'n_estimators':[100,200, 250, 300, 350, 400, 450],
'max_features':[1,2,3,4,5,6,7]}
grid_search = GridSearchCV(RandomForestClassifier(), param_grid, scoring = 'recall',cv = 5,
return_train_score = True)
grid_search.fit(X_train, Y_train)
best_parameters = grid_search.best_params_
print(best_parameters)
print("Test Set Score: {:.2f}".format(grid_search.score(X_test, Y_test)))
#Test random forest with new hyperparameters
#
rfc = RandomForestClassifier(n_estimators = 400, criterion = 'entropy', max_features = 5)#Best parameters combination
rfc.fit(X_train,Y_train)
Y_pred = rfc.predict(X_test)
print("Prediction Accuracy: ", metrics.recall_score(Y_test, Y_pred))
conf_mat = metrics.confusion_matrix(Y_test, Y_pred)
plt.figure(figsize=(8,6))
sns.heatmap(conf_mat,annot=True)
plt.title("Confusion_matrix")
plt.xlabel("Predicted Class")
plt.ylabel("Actual class")
plt.show()
print('Confusion matrix: \n', conf_mat)
print('TP: ', conf_mat[1,1])
print('TN: ', conf_mat[0,0])
print('FP: ', conf_mat[0,1])
print('FN: ', conf_mat[1,0])
################################################################## Gradient Boosting Classifier ####################################
#Implement GBRT
#
gbrt= GradientBoostingClassifier(n_estimators = 200,max_depth=4,learning_rate = 1)# best performer
gbrt.fit(X_train, Y_train)
Y_pred = gbrt.predict(X_test)
print("Prediction Recall: ", metrics.recall_score(Y_test, Y_pred))
conf_mat = metrics.confusion_matrix(Y_test, Y_pred)
plt.figure(figsize=(8,6))
sns.heatmap(conf_mat,annot=True)
plt.title("Confusion_matrix")
plt.xlabel("Predicted Class")
plt.ylabel("Actual class")
plt.show()
print('Confusion matrix: \n', conf_mat)
print('TP: ', conf_mat[1,1])
print('TN: ', conf_mat[0,0])
print('FP: ', conf_mat[0,1])
print('FN: ', conf_mat[1,0])
#Implementing Grid Search CV
#
param_grid = {'max_depth':[1,3,4,5],
'n_estimators':[100,250, 350],
'learning_rate':[0.1,0.5,1]}
grid_search = GridSearchCV(GradientBoostingClassifier(), param_grid, scoring = 'recall',cv = 5,
return_train_score = True)
grid_search.fit(X_train, Y_train)
best_parameters = grid_search.best_params_
print(best_parameters)
print("Test Set Score: {:.2f}".format(grid_search.score(X_test, Y_test)))
#Test gradient boosting classifier with new hyperparameters
#
gbrt= GradientBoostingClassifier(n_estimators = 200,max_depth=4,learning_rate =1)
gbrt.fit(X_train, Y_train)
Y_pred = gbrt.predict(X_test)
print("Prediction Recall: ", metrics.recall_score(Y_test, Y_pred))
conf_mat = metrics.confusion_matrix(Y_test, Y_pred)
plt.figure(figsize=(8,6))
sns.heatmap(conf_mat,annot=True)
plt.title("Confusion_matrix")
plt.xlabel("Predicted Class")
plt.ylabel("Actual class")
plt.show()
print('Confusion matrix: \n', conf_mat)
print('TP: ', conf_mat[1,1])
print('TN: ', conf_mat[0,0])
print('FP: ', conf_mat[0,1])
print('FN: ', conf_mat[1,0])
#Auc RFCR
fpr,tpr,thresholds = roc_curve(Y_test, rfc.decision_function(X_test))
plt.plot(fpr,tpr, label='GBRT')
plt.show()
print("AUC GBRT: {:.3f}".format(roc_auc_score(Y_test, rfc.decision_function(X_test))))
#Auc GBRT
fpr,tpr,thresholds = roc_curve(Y_test, gbrt.decision_function(X_test))
plt.plot(fpr,tpr, label='GBRT')
plt.show()
print("AUC GBRT: {:.3f}".format(roc_auc_score(Y_test, gbrt.decision_function(X_test))))