-
Notifications
You must be signed in to change notification settings - Fork 0
/
Survival_Analysis_V2.0.py
136 lines (113 loc) · 5.97 KB
/
Survival_Analysis_V2.0.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
''' title: "Pre-process the merged dataset"
author: "Raza Mehar"
date: "2024-03-02"
description: Handling missing and duplicate values. Bivariate analyses. Survival analysis'''
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from lifelines import KaplanMeierFitter, CoxPHFitter
data = pd.read_csv("Manning_April_2023.csv")
data.set_index("Person ID", inplace=True)
print("The initial rows of the data are:\n", data.head())
print(data.shape)
# Converting data features to Pandas data type and creating Survival Time feature
data["Start Date"] = pd.to_datetime(data["Start Date"], infer_datetime_format=True, errors="coerce")
data["End Date"] = pd.to_datetime(data["End Date"], infer_datetime_format=True, errors="coerce")
data["Survival Time"] = (data["End Date"] - data["Start Date"]).dt.days
data["Survival Time"] = data["Survival Time"] / 30
# Checking and handling duplicate and missing values
print("\nThe duplicate values are:", data.duplicated().sum())
print("\nThe missing values are:\n", data.isna().sum())
print("The data types of the features are:\n", data.dtypes)
kp = KaplanMeierFitter()
kp.fit(durations=data["Survival Time"], event_observed=data["Left"], label="Kaplan-Meier Survival Curve")
kp.plot_survival_function(linewidth=2, figsize=(12, 6))
plt.xlabel("Time (months)")
plt.xticks(np.arange(0, max(data["Survival Time"]), 30))
plt.ylabel("Survival Probability")
plt.show()
tenure_3 = data[data["Survival Time"] < 36]
tenure_5 = data[(data["Survival Time"] >= 36) & (data["Survival Time"] < 60)]
tenure_10 = data[(data["Survival Time"] >= 60) & (data["Survival Time"] < 120)]
tenure_10above = data[data["Survival Time"] >= 120]
m1_m2 = data[(data["Band"] == "M1") | (data["Band"] == "M2")]
m3 = data[(data["Band"] == "M3")]
p3_p5 = data[(data["Band"] == "P3") | (data["Band"] == "P4") | (data["Band"] == "P5")]
p6_p7 = data[(data["Band"] == "P6") | (data["Band"] == "P7") | (data["Band"] == "T3")]
fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2, figsize=(12, 8), sharey=True, gridspec_kw={'hspace': 0.5})
tenure_3.groupby("Left")["Survival Time"].plot(kind="hist", ax=ax1, title="0 - 3")
tenure_5.groupby("Left")["Survival Time"].plot(kind="hist", ax=ax2, title="3 - 5")
tenure_10.groupby("Left")["Survival Time"].plot(kind="hist", ax=ax3, title="5 - 10")
tenure_10above.groupby("Left")["Survival Time"].plot(kind="hist", ax=ax4, title="10+")
plt.xticks(np.arange(0, max(data["Survival Time"]), 30))
ax1.set_xlabel("Time (months)")
ax2.set_xlabel("Time (months)")
ax3.set_xlabel("Time (months)")
ax4.set_xlabel("Time (months)")
plt.show()
ax = plt.axes()
kp.fit(durations=tenure_3["Survival Time"], event_observed=tenure_3["Left"], label="0 - 3")
kp.plot_survival_function(linewidth=2, figsize=(12, 6), ax=ax)
kp.fit(durations=tenure_5["Survival Time"], event_observed=tenure_5["Left"], label="3 - 5")
kp.plot_survival_function(linewidth=2, figsize=(12, 6), ax=ax)
kp.fit(durations=tenure_10["Survival Time"], event_observed=tenure_10["Left"], label="5 - 10")
kp.plot_survival_function(linewidth=2, figsize=(12, 6), ax=ax)
kp.fit(durations=tenure_10above["Survival Time"], event_observed=tenure_10above["Left"], label="10+")
kp.plot_survival_function(linewidth=2, figsize=(12, 6), ax=ax)
plt.xlabel("Time (months)")
plt.xticks(np.arange(0, max(data["Survival Time"]), 30))
plt.ylabel("Survival Probability")
plt.show()
fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2, figsize=(12, 8), sharey=True, gridspec_kw={'hspace': 0.5})
m1_m2.groupby("Left")["Survival Time"].plot(kind="hist", ax=ax1, title="M1-M2")
m3.groupby("Left")["Survival Time"].plot(kind="hist", ax=ax2, title="M3")
p3_p5.groupby("Left")["Survival Time"].plot(kind="hist", ax=ax3, title="P3-P5")
p6_p7.groupby("Left")["Survival Time"].plot(kind="hist", ax=ax4, title="P6-P7")
plt.xticks(np.arange(0, max(data["Survival Time"]), 30))
ax1.legend(labels=["Not churned", "Churned"])
ax1.set_xlabel("Time (months)")
ax2.set_xlabel("Time (months)")
plt.show()
ax = plt.axes()
kp.fit(durations=m1_m2["Survival Time"], event_observed=m1_m2["Left"], label="M1-M2")
kp.plot_survival_function(linewidth=2, figsize=(12, 6), ax=ax)
kp.fit(durations=m3["Survival Time"], event_observed=m3["Left"], label="M3")
kp.plot_survival_function(linewidth=2, figsize=(12, 6), ax=ax)
kp.fit(durations=p3_p5["Survival Time"], event_observed=p3_p5["Left"], label="P3-P5")
kp.plot_survival_function(linewidth=2, figsize=(12, 6), ax=ax)
kp.fit(durations=p6_p7["Survival Time"], event_observed=p6_p7["Left"], label="P6-P7")
kp.plot_survival_function(linewidth=2, figsize=(12, 6), ax=ax)
plt.xlabel("Time (months)")
plt.xticks(np.arange(0, max(data["Survival Time"]), 30))
plt.ylabel("Survival Probability")
plt.show()
conditions = [
(data["Band"].isin(["M1", "M2"])),
(data["Band"].isin(["M3"])),
(data["Band"].isin(["M4"])),
(data["Band"].isin(["P3", "P4", "P5"])),
(data["Band"].isin(["P6", "P7", "T3"]))
]
choices = ["M1-M2", "M3", "M4", "P3-P5", "P6-T3"]
data["Band_Category"] = np.select(conditions, choices, default='Other')
columns = data[["Band_Category", "Department", "Left"]]
encoded_data = pd.get_dummies(columns, drop_first=True)
encoded_data["Survival Time"] = data["Survival Time"]
cpf = CoxPHFitter()
cpf.fit(encoded_data, duration_col="Survival Time", event_col="Left")
cpf.print_summary(style="ascii")
'''cpf.plot_partial_effects_on_outcome("Gender_Male", [1, 0], plot_baseline=False)
plt.show()'''
data1 = data[data["Left"] == 1]
fig, (ax1, ax2) = plt.subplots(nrows=2, ncols=1, figsize=(14, 8))
sns.boxplot(data=data1, x="Band", y="Survival Time", ax=ax1)
ax1.set_ylabel("")
ax1.set_title("Tenure Distribution by Band for Employees Who Have Left")
sns.boxplot(data=data1, x="Department", y="Survival Time", ax=ax2)
ax2.set_ylabel("") # Set y-axis label
ax2.set_title("Tenure Distribution by Department for Employees Who Have Left")
plt.xticks(rotation=45)
plt.ylabel("")
plt.tight_layout()
plt.show()