Skip to content
/ stats_churn Public

A case study to estimate employee turn over rate using machine learning

2 stars 0 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

YIZHE12/stats_churn

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
EDA_survival_analysis.ipynb
EDA_survival_analysis.ipynb
 
 
README.md
README.md
 
 
To leave or not to leave_ (1).pdf
To leave or not to leave_ (1).pdf
 
 
company_survival.png
company_survival.png
 
 
dept_survival.png
dept_survival.png
 
 
employee_retention.txt
employee_retention.txt
 
 
output_20_0.png
output_20_0.png
 
 
output_23_0.png
output_23_0.png
 
 
output_24_0.png
output_24_0.png
 
 
output_35_1.png
output_35_1.png
 
 
output_36_0.png
output_36_0.png
 
 
output_47_1.png
output_47_1.png
 
 
output_48_1.png
output_48_1.png
 
 
output_49_1.png
output_49_1.png
 
 
output_50_1.png
output_50_1.png
 
 
salary.png
salary.png
 
 
seniority.png
seniority.png
 
 
stat.png
stat.png
 
 
stat2.png
stat2.png
 
 

Repository files navigation

SurvivalAnalysis

NOTE: github's rendering in notebook doesn't always work, if the ipynb file doesn't load, you can see it in https://nbviewer.jupyter.org/github/YIZHE12/SurvivalAnalysis/blob/master/EDA_survival_analysis.ipynb

Background:

You belong to the people analytics team for a food conglomerate. Employee turnover has been rampant for your 10 subsidiaries. The CFO estimates that the cost of replacing an employee is often larger than 100K USD, taking into account the time spent to interview and find a replacement, placement fees, sign-on bonuses and the loss of productivity for several months.

Your team has been tasked with diagnosing why and when employees from your subsidiaries leave. You need a tangible data-driven recommendation for each of the ten Presidents of your subsidiaries. What are your recommendations and why?

Quick look:

This is a survival analysis tasks that I solved using Kaplan Meier plot and Cox Proportional-Hazards Model. There are some data cleaning to do as the datedata is several formats. There are also outliers in the data. For examples, two data points have seniority of 90 years which is not likely as we don't expect someone who have worked for 90 years. For more information, you can have a look at the pdf file in the repo.

The notebook here included all the analysis. The data is the txt file uploaded.

Prerequisites

pandas, numpy, lifelines, matplotlib, seaborn

pip install numpy
pip install pandas
pip install matplotlib
pip install seaborn
pip install lifelines

The notebook:

import os
import pandas as pd
import datetime
import matplotlib.pyplot as plt
import seaborn as sns
import matplotlib
import numpy as np
matplotlib.rcParams.update({'font.size': 20})
data = pd.read_csv('employee_retention.txt', index_col = 'Unnamed: 0')
data.head(5)
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
employee_id company_id dept seniority salary join_date quit_date
0 1001444.0 8 temp_contractor 0 5850.0 2008-01-26 2008-04-25
1 388804.0 8 design 21 191000.0 05.17.2011 2012-03-16
2 407990.0 3 design 9 90000.0 2012-03-26 2015-04-10
3 120657.0 2 engineer 20 298000.0 2013-04-08 2015-01-30
4 1006393.0 1 temp_contractor 0 8509.0 2008-07-20 2008-10-18 
data.info()
<class 'pandas.core.frame.DataFrame'>
Int64Index: 34702 entries, 0 to 34701
Data columns (total 7 columns):
employee_id    34702 non-null float64
company_id     34702 non-null int64
dept           34702 non-null object
seniority      34702 non-null int64
salary         34463 non-null float64
join_date      34702 non-null object
quit_date      23510 non-null object
dtypes: float64(2), int64(2), object(3)
memory usage: 2.1+ MB

There are less data in the quit_date, this is because these people are not quite yet

Some salary data is missing

But maybe for the first survivol analysis we can do without

Notice the date has different formats: 05.17.2011 2012-03-16

Data cleaning

data.join_date = pd.to_datetime(data.join_date)
data.quit_date = pd.to_datetime(data.quit_date)

# data.quit_date = data.quit_date.fillna(value=datetime.date.today())
data.quit_date.max()
Timestamp('2015-12-09 00:00:00')

data.join_date.max()
Timestamp('2015-12-10 00:00:00')

data.employee_id = data.employee_id.astype('int32')
len(data.employee_id.unique()) # check if there is replicate
34702

len(data.company_id.unique()) 
12

data.dept.unique()
array(['temp_contractor', 'design', 'engineer', 'marketing',
       'customer_service', 'data_science', 'sales'], dtype=object)

Does every company has the all these departments?

for i in range(len(data.company_id.unique())):
    print(i+1, data[data.company_id == i+1].dept.unique(), len(data[data.company_id == i+1].dept.unique()))
1 ['temp_contractor' 'customer_service' 'engineer' 'sales' 'data_science'
 'marketing' 'design'] 7
2 ['engineer' 'data_science' 'design' 'temp_contractor' 'sales'
 'customer_service' 'marketing'] 7
3 ['design' 'customer_service' 'data_science' 'sales' 'temp_contractor'
 'marketing' 'engineer'] 7
4 ['temp_contractor' 'data_science' 'marketing' 'customer_service'
 'engineer' 'design' 'sales'] 7
5 ['marketing' 'sales' 'temp_contractor' 'customer_service' 'data_science'
 'design' 'engineer'] 7
6 ['marketing' 'temp_contractor' 'engineer' 'design' 'customer_service'
 'data_science' 'sales'] 7
7 ['data_science' 'design' 'temp_contractor' 'customer_service' 'engineer'
 'marketing' 'sales'] 7
8 ['temp_contractor' 'design' 'customer_service' 'engineer' 'sales'
 'marketing' 'data_science'] 7
9 ['temp_contractor' 'customer_service' 'engineer' 'sales' 'data_science'
 'marketing' 'design'] 7
10 ['data_science' 'temp_contractor' 'marketing' 'customer_service'
 'engineer' 'sales' 'design'] 7
11 ['engineer' 'customer_service' 'marketing' 'data_science'] 4
12 ['data_science' 'engineer' 'customer_service' 'marketing' 'sales' 'design'] 6

How many examples each companies have, and for which department?

count_company = data.groupby('company_id').count()
count_company
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
employee_id dept seniority salary join_date quit_date
company_id
1 9501 9501 9501 9423 9501 5636
2 5220 5220 5220 5178 5220 3204
3 3773 3773 3773 3748 3773 2555
4 3066 3066 3066 3046 3066 2157
5 2749 2749 2749 2734 2749 1977
6 2258 2258 2258 2243 2258 1679
7 2185 2185 2185 2170 2185 1653
8 2026 2026 2026 2011 2026 1558
9 2005 2005 2005 1998 2005 1573
10 1879 1879 1879 1873 1879 1494
11 16 16 16 16 16 12
12 24 24 24 23 24 12

company 11 and 12 have too few data points!

count_department = data.groupby(['company_id', 'dept']).count()
count_department
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
employee_id seniority salary join_date quit_date
company_id dept
1 customer_service 3157 3157 3129 3157 1803
data_science 1079 1079 1070 1079 565
design 499 499 491 499 269
engineer 1568 1568 1552 1568 748
marketing 1085 1085 1075 1085 620
sales 1098 1098 1091 1098 616
temp_contractor 1015 1015 1015 1015 1015
2 customer_service 1548 1548 1530 1548 840
data_science 568 568 562 568 269
design 223 223 223 223 126
engineer 829 829 822 829 384
marketing 541 541 535 541 295
sales 513 513 508 513 292
temp_contractor 998 998 998 998 998
3 customer_service 1010 1010 1000 1010 545
data_science 347 347 345 347 194
design 141 141 141 141 81
engineer 516 516 512 516 292
marketing 372 372 367 372 214
sales 363 363 359 363 205
temp_contractor 1024 1024 1024 1024 1024
4 customer_service 777 777 769 777 415
data_science 279 279 277 279 161
design 107 107 106 107 61
engineer 376 376 375 376 208
marketing 269 269 263 269 157
sales 254 254 252 254 151
temp_contractor 1004 1004 1004 1004 1004
5 customer_service 635 635 631 635 355
data_science 216 216 213 216 114
... ... ... ... ... ... ...
8 data_science 146 146 143 146 80
design 53 53 53 53 24
engineer 191 191 190 191 103
marketing 135 135 132 135 68
sales 137 137 136 137 85
temp_contractor 979 979 979 979 979
9 customer_service 342 342 341 342 186
data_science 134 134 133 134 71
design 60 60 58 60 41
engineer 188 188 185 188 106
marketing 124 124 124 124 62
sales 113 113 113 113 63
temp_contractor 1044 1044 1044 1044 1044
10 customer_service 336 336 333 336 190
data_science 109 109 108 109 52
design 41 41 41 41 23
engineer 172 172 171 172 94
marketing 96 96 96 96 56
sales 111 111 110 111 65
temp_contractor 1014 1014 1014 1014 1014
11 customer_service 6 6 6 6 3
data_science 2 2 2 2 2
engineer 6 6 6 6 5
marketing 2 2 2 2 2
12 customer_service 12 12 11 12 7
data_science 4 4 4 4 2
design 1 1 1 1 0
engineer 4 4 4 4 1
marketing 1 1 1 1 0
sales 2 2 2 2 2

80 rows × 5 columns

fig, ax = plt.subplots(figsize=(20,10))

count_department['employee_id'].unstack().plot(ax=ax, kind = 'bar')
plt.savefig('stat.png')

png

n_company = len(data.company_id.unique())
n_dept= len(data.dept.unique())
companies = data.company_id.unique()
plt.figure(figsize = (40, 30))
for i, company in enumerate(companies):
    plt.subplot(4,3,i+1)
    sns.distplot(data[data.company_id == company].salary.dropna(), norm_hist=False, kde=False, bins=20, \
                 hist_kws={"alpha": 1}).set(xlabel='Salary', ylabel='Count');
    plt.title('company '+str(i+1) + ' salary' )
#     sns.distplot(data[data.company_id == company].salary.dropna(),norm_hist=False)
plt.savefig('salary.png')

png

plt.figure(figsize = (40, 30))
for i, company in enumerate(companies):
    plt.subplot(4,3,i+1)
    sns.distplot(data[data.company_id == company].seniority.dropna(), norm_hist=False, kde=False, bins=20, \
                 hist_kws={"alpha": 1}).set(xlabel='seniority', ylabel='Count');
    plt.title('company '+str(i+1) + ' seniority' )
#     sns.distplot(data[data.company_id == company].salary.dropna(),norm_hist=False)
plt.savefig('seniority.png')

png

Create another copy of data to do numerical EDA

data_num = data.copy()
data_num.head(5)
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
employee_id company_id dept seniority salary join_date quit_date
0 1001444 8 temp_contractor 0 5850.0 2008-01-26 2008-04-25
1 388804 8 design 21 191000.0 2011-05-17 2012-03-16
2 407990 3 design 9 90000.0 2012-03-26 2015-04-10
3 120657 2 engineer 20 298000.0 2013-04-08 2015-01-30
4 1006393 1 temp_contractor 0 8509.0 2008-07-20 2008-10-18 
# data_num.quit_date = data_num.quit_date.fillna(value=datetime.date.today())
data_num = data_num.dropna()
data_num.info()
<class 'pandas.core.frame.DataFrame'>
Int64Index: 23379 entries, 0 to 34701
Data columns (total 7 columns):
employee_id    23379 non-null int32
company_id     23379 non-null int64
dept           23379 non-null object
seniority      23379 non-null int64
salary         23379 non-null float64
join_date      23379 non-null datetime64[ns]
quit_date      23379 non-null datetime64[ns]
dtypes: datetime64[ns](2), float64(1), int32(1), int64(2), object(1)
memory usage: 1.3+ MB

data_num['lasting_data'] = data_num.quit_date - data_num.join_date
data_num['lasting_data'] = data_num['lasting_data'].dt.days
data_num.head(5)
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
employee_id company_id dept seniority salary join_date quit_date lasting_data
0 1001444 8 temp_contractor 0 5850.0 2008-01-26 2008-04-25 90
1 388804 8 design 21 191000.0 2011-05-17 2012-03-16 304
2 407990 3 design 9 90000.0 2012-03-26 2015-04-10 1110
3 120657 2 engineer 20 298000.0 2013-04-08 2015-01-30 662
4 1006393 1 temp_contractor 0 8509.0 2008-07-20 2008-10-18 90 
data_num.company_id = pd.Categorical(data_num.company_id)
data_num.company_id = data_num.company_id.astype('object')
data_num.info()
<class 'pandas.core.frame.DataFrame'>
Int64Index: 23379 entries, 0 to 34701
Data columns (total 8 columns):
employee_id     23379 non-null int32
company_id      23379 non-null object
dept            23379 non-null object
seniority       23379 non-null int64
salary          23379 non-null float64
join_date       23379 non-null datetime64[ns]
quit_date       23379 non-null datetime64[ns]
lasting_data    23379 non-null int64
dtypes: datetime64[ns](2), float64(1), int32(1), int64(2), object(2)
memory usage: 1.5+ MB

g = sns.pairplot(data_num[["company_id", "seniority", "salary", "lasting_data"]], \
                 hue = "company_id")
/home/ubuntu/anaconda3/envs/tensorflow_p36/lib/python3.7/site-packages/scipy/stats/stats.py:1713: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  return np.add.reduce(sorted[indexer] * weights, axis=axis) / sumval
/home/ubuntu/anaconda3/envs/tensorflow_p36/lib/python3.7/site-packages/statsmodels/nonparametric/kde.py:487: RuntimeWarning: invalid value encountered in true_divide
  binned = fast_linbin(X, a, b, gridsize) / (delta * nobs)
/home/ubuntu/anaconda3/envs/tensorflow_p36/lib/python3.7/site-packages/statsmodels/nonparametric/kdetools.py:34: RuntimeWarning: invalid value encountered in double_scalars
  FAC1 = 2*(np.pi*bw/RANGE)**2
/home/ubuntu/anaconda3/envs/tensorflow_p36/lib/python3.7/site-packages/numpy/core/fromnumeric.py:83: RuntimeWarning: invalid value encountered in reduce
  return ufunc.reduce(obj, axis, dtype, out, **passkwargs)

png

g = sns.pairplot(data_num[["dept", "seniority", "salary", "lasting_data"]], \
                 hue = "dept")

png

Survival analysis

data['event'] = pd.isnull(data.quit_date).astype('int8') 
# if there is a quit day, it is not nan, flag 0 - quit
data.quit_date.max()
Timestamp('2015-12-09 00:00:00')

2015-12-10 is the maximum date in the file

data.quit_date = data.quit_date.fillna(value='2015-12-13')
data.quit_date = pd.to_datetime(data.quit_date)
data.head(50)
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
employee_id company_id dept seniority salary join_date quit_date event
0 1001444 8 temp_contractor 0 5850.0 2008-01-26 2008-04-25 0
1 388804 8 design 21 191000.0 2011-05-17 2012-03-16 0
2 407990 3 design 9 90000.0 2012-03-26 2015-04-10 0
3 120657 2 engineer 20 298000.0 2013-04-08 2015-01-30 0
4 1006393 1 temp_contractor 0 8509.0 2008-07-20 2008-10-18 0
5 287530 5 marketing 20 180000.0 2014-06-30 2015-12-13 1
6 561043 3 customer_service 18 119000.0 2012-07-02 2014-03-28 0
7 702479 7 data_science 7 140000.0 2011-12-27 2013-08-30 0
8 545690 10 data_science 16 238000.0 2013-12-23 2015-12-13 1
9 622587 5 sales 28 166000.0 2015-07-01 2015-12-13 1
10 430126 2 data_science 3 77000.0 2015-08-03 2015-12-13 1
11 838072 3 data_science 13 162000.0 2011-10-03 2012-08-10 0
12 205557 8 customer_service 17 109000.0 2013-07-22 2014-07-18 0
13 554514 1 customer_service 4 33000.0 2013-04-15 2015-04-24 0
14 14751 7 design 18 162000.0 2012-04-30 2014-02-14 0
15 602443 3 sales 16 150000.0 2011-09-12 2013-07-19 0
16 488083 1 engineer 8 NaN 2011-06-13 2013-06-07 0
17 1007464 7 temp_contractor 0 7748.0 2009-11-14 2010-02-12 0
18 1002775 3 temp_contractor 0 7424.0 2008-01-14 2008-04-13 0
19 581423 6 marketing 1 35000.0 2012-01-09 2015-06-12 0
20 1000103 5 temp_contractor 0 9684.0 2008-05-18 2008-08-16 0
21 34604 2 design 29 224000.0 2015-09-08 2015-12-13 1
22 1008116 4 temp_contractor 0 9865.0 2010-10-03 2011-01-01 0
23 182278 1 sales 19 179000.0 2011-09-19 2012-11-02 0
24 1003092 2 temp_contractor 0 5459.0 2009-09-23 2009-12-22 0
25 296069 2 engineer 16 308000.0 2012-01-03 2015-12-13 1
26 1007778 7 temp_contractor 0 6749.0 2007-02-14 2007-05-15 0
27 612255 7 customer_service 6 66000.0 2015-03-23 2015-12-13 1
28 28269 2 sales 9 153000.0 2011-08-29 2012-08-03 0
29 904543 2 data_science 17 314000.0 2013-11-25 2015-12-13 1
30 289336 3 design 6 111000.0 2012-12-24 2015-06-26 0
31 591606 1 customer_service 22 123000.0 2015-08-24 2015-12-13 1
32 505031 8 engineer 15 229000.0 2013-09-09 2015-12-13 1
33 1006601 8 temp_contractor 0 9051.0 2008-03-11 2008-06-09 0
34 855236 2 engineer 19 309000.0 2012-01-17 2015-12-13 1
35 543068 5 sales 1 42000.0 2012-08-14 2013-08-02 0
36 282308 4 data_science 14 130000.0 2014-04-11 2015-12-13 1
37 1005290 2 temp_contractor 0 9723.0 2008-09-19 2008-12-18 0
38 643275 3 customer_service 3 21000.0 2011-06-13 2012-06-08 0
39 115980 3 marketing 6 100000.0 2012-04-09 2013-04-26 0
40 1006290 4 temp_contractor 0 9512.0 2008-10-13 2009-01-11 0
41 259298 1 engineer 9 NaN 2011-11-07 2015-10-16 0
42 1007928 10 temp_contractor 0 6538.0 2009-12-28 2010-03-28 0
43 13088 2 customer_service 4 34000.0 2015-09-21 2015-12-13 1
44 1004117 10 temp_contractor 0 8052.0 2007-05-17 2007-08-15 0
45 1002404 10 temp_contractor 0 7998.0 2009-09-13 2009-12-12 0
46 975096 1 customer_service 16 125000.0 2015-04-27 2015-12-13 1
47 432323 7 engineer 20 236000.0 2013-12-02 2015-12-13 1
48 921758 2 engineer 9 216000.0 2014-04-08 2015-01-02 0
49 301501 7 customer_service 29 93000.0 2014-05-27 2015-04-10 0 
data['lasting_days'] = data.quit_date - data.join_date
data['lasting_days'] = data['lasting_days'].dt.days
from lifelines import KaplanMeierFitter
kmf = KaplanMeierFitter()
matplotlib.rcParams.update({'font.size': 10})
fig, axes = plt.subplots(1, 1, figsize=(9, 5))

## Fit the data into the model
kmf.fit(data['lasting_days'] , data['event'], label='Kaplan Meier Estimate')
## Create an estimate
kmf.plot(ci_show=False, ax=axes) 
plt.legend(loc='lower left')
kmf.median_
## ci_show is meant for Confidence interval, since our data set is too tiny, thus i am not showing it.
909.0

png

department_type = data['dept'].unique()
matplotlib.rcParams.update({'font.size': 25})
fig, axes = plt.subplots(1, 1, figsize=(20, 20))
for i, de in enumerate(department_type): 
    
    i1 = (data.dept == de)      ## group i1 , having the pandas series  for the 1st cohort
    ## fit the model for 1st cohort
    kmf.fit(data['lasting_days'][i1] , data['event'][i1], label=de)   
    print(de, ':', kmf.median_)
    kmf.plot(ci_show=False, ax=axes)
    
    plt.yscale('log')
    
plt.savefig('dept_survival.png')
temp_contractor : inf
design : 909.0
engineer : 888.0
marketing : 909.0
customer_service : 888.0
data_science : 902.0
sales : 895.0

png

fig, ax = plt.subplots(1, 1, figsize=(20, 20))
kmf = KaplanMeierFitter()

for name, grouped_df in data.groupby(['company_id']):
    kmf.fit(grouped_df["lasting_days"], grouped_df["event"], label=name)
    print(name, ':', kmf.median_)
    kmf.plot(ax=ax)
    
plt.savefig('company_survival.png')
1 : 895.0
2 : 902.0
3 : 916.0
4 : 902.0
5 : 923.0
6 : 923.0
7 : 923.0
8 : 929.0
9 : 916.0
10 : 937.0
11 : 1217.0
12 : 726.0

png

fig, ax = plt.subplots(1, 1, figsize=(20, 20))
kmf = KaplanMeierFitter()

for name, grouped_df in data.groupby(['company_id', 'dept']):
    kmf.fit(grouped_df["lasting_days"], grouped_df["event"], label=name)
    print(name, ':', kmf.median_)
    kmf.plot(ci_show=False, ax=ax)
    
(1, 'customer_service') : 916.0
(1, 'data_science') : 902.0
(1, 'design') : 881.0
(1, 'engineer') : 846.0
(1, 'marketing') : 929.0
(1, 'sales') : 853.0
(1, 'temp_contractor') : inf
(2, 'customer_service') : 891.0
(2, 'data_science') : 895.0
(2, 'design') : 923.0
(2, 'engineer') : 864.0
(2, 'marketing') : 937.0
(2, 'sales') : 888.0
(2, 'temp_contractor') : inf
(3, 'customer_service') : 888.0
(3, 'data_science') : 825.0
(3, 'design') : 1000.0
(3, 'engineer') : 937.0
(3, 'marketing') : 951.0
(3, 'sales') : 865.0
(3, 'temp_contractor') : inf
(4, 'customer_service') : 825.0
(4, 'data_science') : 1007.0
(4, 'design') : 811.0
(4, 'engineer') : 937.0
(4, 'marketing') : 1014.0
(4, 'sales') : 867.0
(4, 'temp_contractor') : inf
(5, 'customer_service') : 853.0
(5, 'data_science') : 923.0
(5, 'design') : 1027.0
(5, 'engineer') : 1021.0
(5, 'marketing') : 727.0
(5, 'sales') : 929.0
(5, 'temp_contractor') : inf
(6, 'customer_service') : 795.0
(6, 'data_science') : 990.0
(6, 'design') : 1055.0
(6, 'engineer') : 881.0
(6, 'marketing') : 1280.0
(6, 'sales') : 909.0
(6, 'temp_contractor') : inf
(7, 'customer_service') : 923.0
(7, 'data_science') : 888.0
(7, 'design') : 816.0
(7, 'engineer') : 909.0
(7, 'marketing') : 923.0
(7, 'sales') : 1021.0
(7, 'temp_contractor') : inf
(8, 'customer_service') : 902.0
(8, 'data_science') : 923.0
(8, 'design') : 691.0
(8, 'engineer') : 950.0
(8, 'marketing') : 846.0
(8, 'sales') : 965.0
(8, 'temp_contractor') : inf
(9, 'customer_service') : 923.0
(9, 'data_science') : 825.0
(9, 'design') : 909.0
(9, 'engineer') : 993.0
(9, 'marketing') : 846.0
(9, 'sales') : 759.0
(9, 'temp_contractor') : inf
(10, 'customer_service') : 902.0
(10, 'data_science') : 797.0
(10, 'design') : 956.0
(10, 'engineer') : 991.0
(10, 'marketing') : 1007.0
(10, 'sales') : 951.0
(10, 'temp_contractor') : inf
(11, 'customer_service') : 587.0
(11, 'data_science') : inf
(11, 'engineer') : 1217.0
(11, 'marketing') : inf
(12, 'customer_service') : 1014.0
(12, 'data_science') : 881.0
(12, 'design') : 699.0
(12, 'engineer') : 699.0
(12, 'marketing') : 726.0
(12, 'sales') : inf

png

from lifelines import CoxPHFitter
data.company_id = data.company_id.astype('category')
# data.seniority = data.seniority.astype('category')
index = np.where(data.seniority>50)
data = data.drop(data.index[index])
data.head(5)
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
employee_id company_id dept seniority salary join_date quit_date event lasting_days
0 1001444 8 temp_contractor 0 5850.0 2008-01-26 2008-04-25 0 90
1 388804 8 design 21 191000.0 2011-05-17 2012-03-16 0 304
2 407990 3 design 9 90000.0 2012-03-26 2015-04-10 0 1110
3 120657 2 engineer 20 298000.0 2013-04-08 2015-01-30 0 662
4 1006393 1 temp_contractor 0 8509.0 2008-07-20 2008-10-18 0 90 
from sklearn.preprocessing import MinMaxScaler
## Create dummy variables
# df_dummy = pd.get_dummies(data[['event','lasting_days','seniority','salary','company_id']], drop_first=True)
df_dummy = data[['event','lasting_days','seniority','salary','company_id','dept']]
df_dummy = df_dummy.dropna()
max_s = np.max(df_dummy.salary.values)
min_s = np.min(df_dummy.salary.values)
df_dummy.salary = (df_dummy.salary.values - min_s)/(max_s-min_s)

max_se = np.max(df_dummy.seniority.values)
min_se = np.min(df_dummy.seniority.values)
df_dummy.seniority = (df_dummy.seniority.values - min_se)/(max_se-min_se)


df_dummy.head()
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
event lasting_days seniority salary company_id dept
0 0 90 0.000000 0.002109 8 temp_contractor
1 0 304 0.724138 0.461538 8 design
2 0 1110 0.310345 0.210918 3 design
3 0 662 0.689655 0.727047 2 engineer
4 0 90 0.000000 0.008707 1 temp_contractor 
# Using Cox Proportional Hazards model
for i in range(12):
    data_test = df_dummy[df_dummy.company_id == i+1]
    data_test = data_test.drop(columns = ['company_id','dept'])   
    cph = CoxPHFitter()   ## Instantiate the class to create a cph object
    cph.fit(data_test, 'lasting_days', event_col='event')   ## Fit the data to train the model
    cph.print_summary()    ## HAve a look at the significance of the features
<lifelines.CoxPHFitter: fitted with 9422 observations, 5596 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 9422
  number of events = 3826
partial log-likelihood = -31068.96
  time fit was run = 2019-07-19 12:58:06 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.02      0.98      0.07           -0.16            0.13                0.86                1.13
salary     0.12      1.13      0.09           -0.06            0.31                0.94                1.36

              z    p  -log2(p)
seniority -0.21 0.83      0.27
salary     1.31 0.19      2.39
---
Concordance = 0.51
Log-likelihood ratio test = 2.27 on 2 df, -log2(p)=1.64
<lifelines.CoxPHFitter: fitted with 5178 observations, 3179 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 5178
  number of events = 1999
partial log-likelihood = -14881.37
  time fit was run = 2019-07-19 12:58:07 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.19      1.21      0.10           -0.01            0.39                0.99                1.47
salary     0.04      1.04      0.13           -0.21            0.29                0.81                1.34

             z    p  -log2(p)
seniority 1.89 0.06      4.08
salary    0.32 0.75      0.42
---
Concordance = 0.53
Log-likelihood ratio test = 7.12 on 2 df, -log2(p)=5.13
<lifelines.CoxPHFitter: fitted with 3748 observations, 2543 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 3748
  number of events = 1205
partial log-likelihood = -8498.81
  time fit was run = 2019-07-19 12:58:07 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.19      1.21      0.13           -0.07            0.45                0.93                1.57
salary    -0.07      0.93      0.22           -0.49            0.36                0.61                1.43

              z    p  -log2(p)
seniority  1.44 0.15      2.73
salary    -0.32 0.75      0.41
---
Concordance = 0.54
Log-likelihood ratio test = 2.64 on 2 df, -log2(p)=1.91
<lifelines.CoxPHFitter: fitted with 3046 observations, 2146 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 3046
  number of events = 900
partial log-likelihood = -6067.60
  time fit was run = 2019-07-19 12:58:07 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.51      1.67      0.15            0.22            0.81                1.25                2.25
salary    -0.29      0.75      0.25           -0.77            0.19                0.46                1.21

              z      p  -log2(p)
seniority  3.43 <0.005     10.67
salary    -1.18   0.24      2.07
---
Concordance = 0.54
Log-likelihood ratio test = 12.81 on 2 df, -log2(p)=9.24
<lifelines.CoxPHFitter: fitted with 2734 observations, 1970 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 2734
  number of events = 764
partial log-likelihood = -5033.96
  time fit was run = 2019-07-19 12:58:07 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.14      1.15      0.17           -0.18            0.47                0.83                1.60
salary    -0.26      0.77      0.28           -0.81            0.28                0.45                1.33

              z    p  -log2(p)
seniority  0.86 0.39      1.35
salary    -0.94 0.35      1.53
---
Concordance = 0.51
Log-likelihood ratio test = 0.99 on 2 df, -log2(p)=0.71
<lifelines.CoxPHFitter: fitted with 2243 observations, 1670 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 2243
  number of events = 573
partial log-likelihood = -3607.36
  time fit was run = 2019-07-19 12:58:08 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.26      1.30      0.18           -0.09            0.62                0.91                1.85
salary    -0.04      0.96      0.30           -0.63            0.55                0.53                1.74

              z    p  -log2(p)
seniority  1.46 0.14      2.80
salary    -0.12 0.90      0.15
---
Concordance = 0.58
Log-likelihood ratio test = 3.08 on 2 df, -log2(p)=2.22
<lifelines.CoxPHFitter: fitted with 2170 observations, 1644 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 2170
  number of events = 526
partial log-likelihood = -3264.59
  time fit was run = 2019-07-19 12:58:08 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.32      1.37      0.20           -0.07            0.70                0.93                2.02
salary     0.18      1.19      0.32           -0.44            0.80                0.64                2.22

             z    p  -log2(p)
seniority 1.60 0.11      3.20
salary    0.56 0.58      0.79
---
Concordance = 0.58
Log-likelihood ratio test = 6.29 on 2 df, -log2(p)=4.54
<lifelines.CoxPHFitter: fitted with 2011 observations, 1547 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 2011
  number of events = 464
partial log-likelihood = -2813.20
  time fit was run = 2019-07-19 12:58:08 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.32      1.38      0.21           -0.09            0.73                0.92                2.07
salary     0.33      1.39      0.34           -0.34            1.00                0.71                2.71

             z    p  -log2(p)
seniority 1.54 0.12      3.03
salary    0.96 0.34      1.57
---
Concordance = 0.58
Log-likelihood ratio test = 8.36 on 2 df, -log2(p)=6.03
<lifelines.CoxPHFitter: fitted with 1998 observations, 1568 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1998
  number of events = 430
partial log-likelihood = -2617.25
  time fit was run = 2019-07-19 12:58:08 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.14      1.15      0.22           -0.29            0.57                0.75                1.77
salary     0.47      1.61      0.35           -0.22            1.17                0.80                3.21

             z    p  -log2(p)
seniority 0.63 0.53      0.92
salary    1.34 0.18      2.48
---
Concordance = 0.60
Log-likelihood ratio test = 5.85 on 2 df, -log2(p)=4.22
<lifelines.CoxPHFitter: fitted with 1872 observations, 1490 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1872
  number of events = 382
partial log-likelihood = -2269.34
  time fit was run = 2019-07-19 12:58:08 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.31      1.36      0.24           -0.17            0.78                0.85                2.18
salary    -0.01      0.99      0.39           -0.77            0.75                0.46                2.12

              z    p  -log2(p)
seniority  1.28 0.20      2.31
salary    -0.02 0.98      0.02
---
Concordance = 0.59
Log-likelihood ratio test = 2.74 on 2 df, -log2(p)=1.97
<lifelines.CoxPHFitter: fitted with 16 observations, 12 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 16
  number of events = 4
partial log-likelihood = -0.00
  time fit was run = 2019-07-19 12:58:09 UTC

---
             coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority   86.63  4.20e+37    659.33        -1205.64         1378.90                0.00                 inf
salary    -257.87      0.00   1823.05        -3830.98         3315.24                0.00                 inf

              z    p  -log2(p)
seniority  0.13 0.90      0.16
salary    -0.14 0.89      0.17
---
Concordance = 1.00
Log-likelihood ratio test = 10.48 on 2 df, -log2(p)=7.56
<lifelines.CoxPHFitter: fitted with 23 observations, 12 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 23
  number of events = 11
partial log-likelihood = -20.07
  time fit was run = 2019-07-19 12:58:09 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -2.03      0.13      3.29           -8.48            4.42                0.00               82.88
salary     3.75     42.71      6.65           -9.28           16.79                0.00            1.96e+07

              z    p  -log2(p)
seniority -0.62 0.54      0.90
salary     0.56 0.57      0.80
---
Concordance = 0.55
Log-likelihood ratio test = 0.39 on 2 df, -log2(p)=0.28


/home/ubuntu/anaconda3/envs/tensorflow_p36/lib/python3.7/site-packages/lifelines/fitters/coxph_fitter.py:561: ConvergenceWarning: Newton-Rhapson failed to converge sufficiently in 50 steps.
  warnings.warn("Newton-Rhapson failed to converge sufficiently in %d steps." % max_steps, ConvergenceWarning)

depts = data.dept.unique()
depts
array(['temp_contractor', 'design', 'engineer', 'marketing',
       'customer_service', 'data_science', 'sales'], dtype=object)

data_test = df_dummy[df_dummy.dept == 'temp_contractor']
data_test = data_test.drop(columns = ['company_id','dept']) 
data_test.head()
<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } 
.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}
</style>
event lasting_days seniority salary
0 0 90 0.0 0.002109
4 0 90 0.0 0.008707
17 0 90 0.0 0.006819
18 0 90 0.0 0.006015
20 0 90 0.0 0.011623 
# Using Cox Proportional Hazards model
for i,de in enumerate(depts):
    if i >1:
        data_test = df_dummy[df_dummy.dept == de]
        data_test = data_test.drop(columns = ['company_id','dept'])   
        cph = CoxPHFitter()   ## Instantiate the class to create a cph object
        cph.fit(data_test, 'lasting_days', event_col='event')   ## Fit the data to train the model
        print(de)
        print('*'*20)
        cph.print_summary()    ## HAve a look at the significance of the features
engineer
********************
<lifelines.CoxPHFitter: fitted with 4568 observations, 2338 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 4568
  number of events = 2230
partial log-likelihood = -16747.83
  time fit was run = 2019-07-19 01:23:11 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.10      0.91      0.13           -0.35            0.15                0.71                1.16
salary     0.03      1.03      0.18           -0.32            0.38                0.72                1.46

              z    p  -log2(p)
seniority -0.77 0.44      1.17
salary     0.16 0.87      0.19
---
Concordance = 0.51
Log-likelihood ratio test = 1.15 on 2 df, -log2(p)=0.83
marketing
********************
<lifelines.CoxPHFitter: fitted with 3132 observations, 1765 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 3132
  number of events = 1367
partial log-likelihood = -9762.54
  time fit was run = 2019-07-19 01:23:11 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.03      1.03      0.17           -0.30            0.37                0.74                1.44
salary    -0.17      0.84      0.36           -0.87            0.53                0.42                1.70

              z    p  -log2(p)
seniority  0.18 0.85      0.23
salary    -0.48 0.63      0.67
---
Concordance = 0.51
Log-likelihood ratio test = 0.38 on 2 df, -log2(p)=0.27
customer_service
********************
<lifelines.CoxPHFitter: fitted with 9089 observations, 5043 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 9089
  number of events = 4046
partial log-likelihood = -33184.68
  time fit was run = 2019-07-19 01:23:12 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.20      1.23      0.10            0.01            0.40                1.01                1.49
salary    -0.77      0.46      0.35           -1.46           -0.09                0.23                0.92

              z    p  -log2(p)
seniority  2.07 0.04      4.71
salary    -2.21 0.03      5.22
---
Concordance = 0.51
Log-likelihood ratio test = 5.11 on 2 df, -log2(p)=3.68
data_science
********************
<lifelines.CoxPHFitter: fitted with 3157 observations, 1663 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 3157
  number of events = 1494
partial log-likelihood = -10679.77
  time fit was run = 2019-07-19 01:23:12 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.03      1.03      0.16           -0.28            0.33                0.75                1.40
salary     0.06      1.06      0.22           -0.37            0.49                0.69                1.63

             z    p  -log2(p)
seniority 0.17 0.87      0.21
salary    0.26 0.79      0.34
---
Concordance = 0.51
Log-likelihood ratio test = 0.47 on 2 df, -log2(p)=0.34
sales
********************
<lifelines.CoxPHFitter: fitted with 3148 observations, 1798 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 3148
  number of events = 1350
partial log-likelihood = -9595.81
  time fit was run = 2019-07-19 01:23:13 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.16      0.85      0.17           -0.49            0.17                0.61                1.19
salary     0.54      1.71      0.35           -0.16            1.23                0.85                3.43

              z    p  -log2(p)
seniority -0.95 0.34      1.54
salary     1.51 0.13      2.94
---
Concordance = 0.52
Log-likelihood ratio test = 2.55 on 2 df, -log2(p)=1.84

# Using Cox Proportional Hazards model
for i in range(10):
    df_dummy2 = df_dummy[df_dummy.company_id == i+1]
    df_dummy2 = df_dummy2.drop(columns = ['company_id'])   
    
# Using Cox Proportional Hazards model
    for j,de in enumerate(depts):
        if j >1:
            data_test = df_dummy2[df_dummy2.dept == de]
            data_test = data_test.drop(columns = ['dept'])   
            cph = CoxPHFitter()   ## Instantiate the class to create a cph object
            cph.fit(data_test, 'lasting_days', event_col='event')   ## Fit the data to train the model
            print("company", i+1, de)
            print('*'*20)
            cph.print_summary()    ## HAve a look at the significance of the features
company 1 engineer
********************
<lifelines.CoxPHFitter: fitted with 1552 observations, 738 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1552
  number of events = 814
partial log-likelihood = -5205.78
  time fit was run = 2019-07-19 01:24:34 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.29      1.34      0.24           -0.18            0.76                0.84                2.14
salary    -0.53      0.59      0.32           -1.15            0.10                0.32                1.10

              z    p  -log2(p)
seniority  1.22 0.22      2.18
salary    -1.66 0.10      3.35
---
Concordance = 0.53
Log-likelihood ratio test = 2.89 on 2 df, -log2(p)=2.08
company 1 marketing
********************
<lifelines.CoxPHFitter: fitted with 1074 observations, 613 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1074
  number of events = 461
partial log-likelihood = -2791.80
  time fit was run = 2019-07-19 01:24:34 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.13      0.88      0.31           -0.74            0.47                0.48                1.61
salary     0.14      1.15      0.63           -1.10            1.38                0.33                3.97

              z    p  -log2(p)
seniority -0.43 0.67      0.58
salary     0.22 0.83      0.27
---
Concordance = 0.51
Log-likelihood ratio test = 0.26 on 2 df, -log2(p)=0.19
company 1 customer_service
********************
<lifelines.CoxPHFitter: fitted with 3129 observations, 1791 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 3129
  number of events = 1338
partial log-likelihood = -9540.69
  time fit was run = 2019-07-19 01:24:34 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.00      1.00      0.19           -0.38            0.38                0.69                1.46
salary    -0.21      0.81      0.64           -1.46            1.05                0.23                2.86

              z    p  -log2(p)
seniority -0.00 1.00      0.00
salary    -0.32 0.75      0.42
---
Concordance = 0.51
Log-likelihood ratio test = 0.40 on 2 df, -log2(p)=0.29
company 1 data_science
********************
<lifelines.CoxPHFitter: fitted with 1070 observations, 562 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1070
  number of events = 508
partial log-likelihood = -3089.21
  time fit was run = 2019-07-19 01:24:34 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.36      1.44      0.29           -0.21            0.93                0.81                2.54
salary    -0.01      0.99      0.41           -0.81            0.79                0.44                2.19

              z    p  -log2(p)
seniority  1.25 0.21      2.25
salary    -0.03 0.97      0.04
---
Concordance = 0.54
Log-likelihood ratio test = 5.10 on 2 df, -log2(p)=3.68
company 1 sales
********************
<lifelines.CoxPHFitter: fitted with 1091 observations, 612 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1091
  number of events = 479
partial log-likelihood = -2914.55
  time fit was run = 2019-07-19 01:24:35 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.69      0.50      0.32           -1.32           -0.05                0.27                0.95
salary     1.20      3.33      0.63           -0.04            2.45                0.96               11.53

              z    p  -log2(p)
seniority -2.12 0.03      4.87
salary     1.90 0.06      4.12
---
Concordance = 0.53
Log-likelihood ratio test = 4.54 on 2 df, -log2(p)=3.28
company 2 engineer
********************
<lifelines.CoxPHFitter: fitted with 822 observations, 380 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 822
  number of events = 442
partial log-likelihood = -2599.58
  time fit was run = 2019-07-19 01:24:35 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.02      0.98      0.31           -0.62            0.58                0.54                1.79
salary     0.11      1.12      0.41           -0.70            0.92                0.50                2.52

              z    p  -log2(p)
seniority -0.06 0.95      0.07
salary     0.27 0.79      0.34
---
Concordance = 0.51
Log-likelihood ratio test = 0.16 on 2 df, -log2(p)=0.12
company 2 marketing
********************
<lifelines.CoxPHFitter: fitted with 535 observations, 291 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 535
  number of events = 244
partial log-likelihood = -1324.46
  time fit was run = 2019-07-19 01:24:35 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.15      1.16      0.47           -0.77            1.06                0.46                2.90
salary    -0.52      0.59      0.92           -2.32            1.28                0.10                3.60

              z    p  -log2(p)
seniority  0.31 0.76      0.41
salary    -0.57 0.57      0.81
---
Concordance = 0.51
Log-likelihood ratio test = 0.45 on 2 df, -log2(p)=0.32
company 2 customer_service
********************
<lifelines.CoxPHFitter: fitted with 1530 observations, 828 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1530
  number of events = 702
partial log-likelihood = -4467.66
  time fit was run = 2019-07-19 01:24:35 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.23      1.26      0.25           -0.26            0.73                0.77                2.07
salary    -0.91      0.40      0.85           -2.58            0.77                0.08                2.15

              z    p  -log2(p)
seniority  0.92 0.36      1.48
salary    -1.06 0.29      1.80
---
Concordance = 0.51
Log-likelihood ratio test = 1.13 on 2 df, -log2(p)=0.82
company 2 data_science
********************
<lifelines.CoxPHFitter: fitted with 562 observations, 265 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 562
  number of events = 297
partial log-likelihood = -1617.21
  time fit was run = 2019-07-19 01:24:35 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.40      1.49      0.41           -0.41            1.21                0.66                3.34
salary    -0.55      0.58      0.57           -1.66            0.57                0.19                1.76

              z    p  -log2(p)
seniority  0.97 0.33      1.58
salary    -0.96 0.34      1.57
---
Concordance = 0.50
Log-likelihood ratio test = 1.00 on 2 df, -log2(p)=0.72
company 2 sales
********************
<lifelines.CoxPHFitter: fitted with 508 observations, 291 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 508
  number of events = 217
partial log-likelihood = -1144.75
  time fit was run = 2019-07-19 01:24:35 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.11      0.89      0.42           -0.94            0.72                0.39                2.05
salary     0.66      1.93      0.90           -1.10            2.41                0.33               11.16

              z    p  -log2(p)
seniority -0.27 0.79      0.34
salary     0.73 0.46      1.11
---
Concordance = 0.52
Log-likelihood ratio test = 0.95 on 2 df, -log2(p)=0.69
company 3 engineer
********************
<lifelines.CoxPHFitter: fitted with 512 observations, 291 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 512
  number of events = 221
partial log-likelihood = -1176.37
  time fit was run = 2019-07-19 01:24:35 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.37      1.45      0.48           -0.57            1.31                0.57                3.72
salary    -1.21      0.30      0.81           -2.79            0.37                0.06                1.45

              z    p  -log2(p)
seniority  0.78 0.44      1.20
salary    -1.50 0.13      2.90
---
Concordance = 0.53
Log-likelihood ratio test = 3.29 on 2 df, -log2(p)=2.37
company 3 marketing
********************
<lifelines.CoxPHFitter: fitted with 367 observations, 213 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 367
  number of events = 154
partial log-likelihood = -771.02
  time fit was run = 2019-07-19 01:24:35 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.36      0.70      0.59           -1.51            0.79                0.22                2.20
salary     0.26      1.29      1.42           -2.53            3.04                0.08               20.95

              z    p  -log2(p)
seniority -0.62 0.54      0.89
salary     0.18 0.86      0.22
---
Concordance = 0.51
Log-likelihood ratio test = 0.95 on 2 df, -log2(p)=0.69
company 3 customer_service
********************
<lifelines.CoxPHFitter: fitted with 1000 observations, 538 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1000
  number of events = 462
partial log-likelihood = -2798.21
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.56      1.75      0.34           -0.11            1.23                0.90                3.41
salary    -3.27      0.04      1.38           -5.98           -0.56                0.00                0.57

              z    p  -log2(p)
seniority  1.65 0.10      3.33
salary    -2.37 0.02      5.79
---
Concordance = 0.52
Log-likelihood ratio test = 6.34 on 2 df, -log2(p)=4.57
company 3 data_science
********************
<lifelines.CoxPHFitter: fitted with 345 observations, 193 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 345
  number of events = 152
partial log-likelihood = -756.00
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.18      0.83      0.57           -1.31            0.94                0.27                2.56
salary     0.48      1.62      0.94           -1.36            2.33                0.26               10.27

              z    p  -log2(p)
seniority -0.32 0.75      0.42
salary     0.51 0.61      0.72
---
Concordance = 0.48
Log-likelihood ratio test = 0.30 on 2 df, -log2(p)=0.22
company 3 sales
********************
<lifelines.CoxPHFitter: fitted with 359 observations, 203 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 359
  number of events = 156
partial log-likelihood = -762.34
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.91      2.50      0.54           -0.15            1.98                0.86                7.25
salary    -0.42      0.66      1.33           -3.03            2.19                0.05                8.92

              z    p  -log2(p)
seniority  1.68 0.09      3.44
salary    -0.32 0.75      0.41
---
Concordance = 0.56
Log-likelihood ratio test = 6.86 on 2 df, -log2(p)=4.95
company 4 engineer
********************
<lifelines.CoxPHFitter: fitted with 375 observations, 208 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 375
  number of events = 167
partial log-likelihood = -845.44
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.29      1.34      0.50           -0.70            1.28                0.50                3.59
salary    -0.27      0.76      0.90           -2.04            1.49                0.13                4.46

              z    p  -log2(p)
seniority  0.58 0.56      0.82
salary    -0.30 0.76      0.39
---
Concordance = 0.50
Log-likelihood ratio test = 0.42 on 2 df, -log2(p)=0.30
company 4 marketing
********************
<lifelines.CoxPHFitter: fitted with 263 observations, 153 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 263
  number of events = 110
partial log-likelihood = -513.20
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.99      2.69      0.67           -0.31            2.29                0.73                9.92
salary    -2.55      0.08      1.70           -5.89            0.79                0.00                2.21

              z    p  -log2(p)
seniority  1.49 0.14      2.88
salary    -1.50 0.13      2.89
---
Concordance = 0.54
Log-likelihood ratio test = 2.42 on 2 df, -log2(p)=1.75
company 4 customer_service
********************
<lifelines.CoxPHFitter: fitted with 769 observations, 412 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 769
  number of events = 357
partial log-likelihood = -2049.08
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.70      2.02      0.35            0.01            1.40                1.01                4.04
salary    -1.75      0.17      1.48           -4.65            1.15                0.01                3.15

              z    p  -log2(p)
seniority  1.99 0.05      4.41
salary    -1.18 0.24      2.08
---
Concordance = 0.51
Log-likelihood ratio test = 4.57 on 2 df, -log2(p)=3.30
company 4 data_science
********************
<lifelines.CoxPHFitter: fitted with 277 observations, 159 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 277
  number of events = 118
partial log-likelihood = -546.75
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.24      1.27      0.67           -1.06            1.55                0.34                4.70
salary    -0.08      0.93      1.12           -2.27            2.11                0.10                8.26

              z    p  -log2(p)
seniority  0.36 0.72      0.48
salary    -0.07 0.95      0.08
---
Concordance = 0.50
Log-likelihood ratio test = 0.31 on 2 df, -log2(p)=0.22
company 4 sales
********************
<lifelines.CoxPHFitter: fitted with 252 observations, 150 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 252
  number of events = 102
partial log-likelihood = -467.89
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.87      0.42      0.69           -2.22            0.47                0.11                1.61
salary     2.31     10.04      1.76           -1.14            5.75                0.32              314.44

              z    p  -log2(p)
seniority -1.27 0.20      2.29
salary     1.31 0.19      2.40
---
Concordance = 0.54
Log-likelihood ratio test = 1.80 on 2 df, -log2(p)=1.30
company 5 engineer
********************
<lifelines.CoxPHFitter: fitted with 311 observations, 178 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 311
  number of events = 133
partial log-likelihood = -645.66
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  1.04      2.84      0.61           -0.15            2.23                0.86                9.34
salary    -2.80      0.06      1.08           -4.92           -0.67                0.01                0.51

              z    p  -log2(p)
seniority  1.72 0.09      3.55
salary    -2.58 0.01      6.67
---
Concordance = 0.55
Log-likelihood ratio test = 7.65 on 2 df, -log2(p)=5.52
company 5 marketing
********************
<lifelines.CoxPHFitter: fitted with 224 observations, 114 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 224
  number of events = 110
partial log-likelihood = -496.74
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.87      0.42      0.72           -2.27            0.53                0.10                1.71
salary     1.84      6.29      1.66           -1.42            5.10                0.24              163.90

              z    p  -log2(p)
seniority -1.21 0.23      2.15
salary     1.11 0.27      1.89
---
Concordance = 0.54
Log-likelihood ratio test = 1.50 on 2 df, -log2(p)=1.08
company 5 customer_service
********************
<lifelines.CoxPHFitter: fitted with 631 observations, 353 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 631
  number of events = 278
partial log-likelihood = -1545.96
  time fit was run = 2019-07-19 01:24:36 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.07      1.07      0.42           -0.77            0.90                0.47                2.45
salary    -1.14      0.32      1.76           -4.59            2.31                0.01               10.09

              z    p  -log2(p)
seniority  0.15 0.88      0.19
salary    -0.65 0.52      0.95
---
Concordance = 0.51
Log-likelihood ratio test = 1.03 on 2 df, -log2(p)=0.74
company 5 data_science
********************
<lifelines.CoxPHFitter: fitted with 213 observations, 113 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 213
  number of events = 100
partial log-likelihood = -440.53
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.51      0.60      0.69           -1.85            0.84                0.16                2.31
salary     0.21      1.24      1.16           -2.06            2.48                0.13               11.92

              z    p  -log2(p)
seniority -0.74 0.46      1.12
salary     0.18 0.86      0.23
---
Concordance = 0.50
Log-likelihood ratio test = 1.10 on 2 df, -log2(p)=0.79
company 5 sales
********************
<lifelines.CoxPHFitter: fitted with 254 observations, 148 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 254
  number of events = 106
partial log-likelihood = -483.46
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.14      1.15      0.63           -1.09            1.38                0.33                3.96
salary    -0.42      0.66      1.62           -3.59            2.75                0.03               15.60

              z    p  -log2(p)
seniority  0.22 0.82      0.28
salary    -0.26 0.79      0.33
---
Concordance = 0.51
Log-likelihood ratio test = 0.07 on 2 df, -log2(p)=0.05
company 6 engineer
********************
<lifelines.CoxPHFitter: fitted with 218 observations, 115 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 218
  number of events = 103
partial log-likelihood = -453.19
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.04      1.04      0.67           -1.27            1.35                0.28                3.86
salary    -0.00      1.00      1.10           -2.16            2.15                0.12                8.59

              z    p  -log2(p)
seniority  0.06 0.95      0.07
salary    -0.00 1.00      0.00
---
Concordance = 0.50
Log-likelihood ratio test = 0.01 on 2 df, -log2(p)=0.01
company 6 marketing
********************
<lifelines.CoxPHFitter: fitted with 174 observations, 117 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 174
  number of events = 57
partial log-likelihood = -252.19
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.64      1.90      0.90           -1.12            2.41                0.33               11.09
salary    -1.18      0.31      2.28           -5.65            3.28                0.00               26.66

              z    p  -log2(p)
seniority  0.72 0.47      1.08
salary    -0.52 0.60      0.73
---
Concordance = 0.57
Log-likelihood ratio test = 0.54 on 2 df, -log2(p)=0.39
company 6 customer_service
********************
<lifelines.CoxPHFitter: fitted with 495 observations, 257 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 495
  number of events = 238
partial log-likelihood = -1260.17
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.13      1.14      0.44           -0.73            1.00                0.48                2.73
salary     0.22      1.24      1.90           -3.52            3.95                0.03               51.94

             z    p  -log2(p)
seniority 0.30 0.76      0.39
salary    0.11 0.91      0.14
---
Concordance = 0.53
Log-likelihood ratio test = 0.63 on 2 df, -log2(p)=0.45
company 6 data_science
********************
<lifelines.CoxPHFitter: fitted with 151 observations, 84 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 151
  number of events = 67
partial log-likelihood = -271.53
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -1.33      0.26      0.82           -2.94            0.28                0.05                1.32
salary     0.79      2.20      1.39           -1.94            3.51                0.14               33.46

              z    p  -log2(p)
seniority -1.62 0.10      3.26
salary     0.57 0.57      0.81
---
Concordance = 0.57
Log-likelihood ratio test = 4.91 on 2 df, -log2(p)=3.54
company 6 sales
********************
<lifelines.CoxPHFitter: fitted with 161 observations, 85 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 161
  number of events = 76
partial log-likelihood = -318.48
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -1.72      0.18      0.79           -3.26           -0.18                0.04                0.84
salary     3.83     46.18      1.93            0.05            7.61                1.05             2025.76

              z    p  -log2(p)
seniority -2.18 0.03      5.11
salary     1.99 0.05      4.41
---
Concordance = 0.59
Log-likelihood ratio test = 4.99 on 2 df, -log2(p)=3.60
company 7 engineer
********************
<lifelines.CoxPHFitter: fitted with 223 observations, 123 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 223
  number of events = 100
partial log-likelihood = -445.24
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.48      0.62      0.69           -1.83            0.88                0.16                2.40
salary     1.23      3.41      1.11           -0.95            3.40                0.39               30.01

              z    p  -log2(p)
seniority -0.69 0.49      1.03
salary     1.10 0.27      1.89
---
Concordance = 0.54
Log-likelihood ratio test = 1.39 on 2 df, -log2(p)=1.00
company 7 marketing
********************
<lifelines.CoxPHFitter: fitted with 140 observations, 77 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 140
  number of events = 63
partial log-likelihood = -243.90
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.54      1.72      0.91           -1.25            2.33                0.29               10.29
salary    -0.92      0.40      2.05           -4.94            3.10                0.01               22.23

              z    p  -log2(p)
seniority  0.59 0.55      0.86
salary    -0.45 0.65      0.61
---
Concordance = 0.54
Log-likelihood ratio test = 0.35 on 2 df, -log2(p)=0.25
company 7 customer_service
********************
<lifelines.CoxPHFitter: fitted with 466 observations, 266 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 466
  number of events = 200
partial log-likelihood = -1045.25
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.07      0.93      0.47           -0.99            0.84                0.37                2.32
salary     0.67      1.96      1.94           -3.12            4.47                0.04               87.26

              z    p  -log2(p)
seniority -0.16 0.87      0.19
salary     0.35 0.73      0.46
---
Concordance = 0.52
Log-likelihood ratio test = 0.17 on 2 df, -log2(p)=0.13
company 7 data_science
********************
<lifelines.CoxPHFitter: fitted with 149 observations, 84 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 149
  number of events = 65
partial log-likelihood = -270.08
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.57      1.77      0.91           -1.21            2.35                0.30               10.52
salary    -1.51      0.22      1.47           -4.39            1.37                0.01                3.95

              z    p  -log2(p)
seniority  0.63 0.53      0.92
salary    -1.03 0.30      1.72
---
Concordance = 0.53
Log-likelihood ratio test = 1.34 on 2 df, -log2(p)=0.96
company 7 sales
********************
<lifelines.CoxPHFitter: fitted with 162 observations, 96 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 162
  number of events = 66
partial log-likelihood = -269.08
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  1.29      3.64      0.85           -0.37            2.95                0.69               19.10
salary    -3.24      0.04      2.09           -7.33            0.84                0.00                2.32

              z    p  -log2(p)
seniority  1.53 0.13      2.98
salary    -1.56 0.12      3.06
---
Concordance = 0.53
Log-likelihood ratio test = 2.57 on 2 df, -log2(p)=1.85
company 8 engineer
********************
<lifelines.CoxPHFitter: fitted with 190 observations, 102 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 190
  number of events = 88
partial log-likelihood = -379.28
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.88      0.41      0.73           -2.32            0.55                0.10                1.73
salary     2.37     10.67      1.32           -0.23            4.96                0.80              143.00

              z    p  -log2(p)
seniority -1.21 0.23      2.14
salary     1.79 0.07      3.76
---
Concordance = 0.55
Log-likelihood ratio test = 3.41 on 2 df, -log2(p)=2.46
company 8 marketing
********************
<lifelines.CoxPHFitter: fitted with 132 observations, 67 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 132
  number of events = 65
partial log-likelihood = -249.50
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.28      1.33      0.96           -1.60            2.16                0.20                8.70
salary     1.67      5.32      2.22           -2.67            6.02                0.07              409.69

             z    p  -log2(p)
seniority 0.30 0.77      0.38
salary    0.75 0.45      1.15
---
Concordance = 0.54
Log-likelihood ratio test = 3.83 on 2 df, -log2(p)=2.76
company 8 customer_service
********************
<lifelines.CoxPHFitter: fitted with 378 observations, 213 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 378
  number of events = 165
partial log-likelihood = -833.58
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.07      1.07      0.51           -0.92            1.06                0.40                2.88
salary     0.42      1.52      2.26           -4.01            4.84                0.02              127.00

             z    p  -log2(p)
seniority 0.13 0.89      0.16
salary    0.19 0.85      0.23
---
Concordance = 0.50
Log-likelihood ratio test = 0.34 on 2 df, -log2(p)=0.24
company 8 data_science
********************
<lifelines.CoxPHFitter: fitted with 143 observations, 78 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 143
  number of events = 65
partial log-likelihood = -261.75
  time fit was run = 2019-07-19 01:24:37 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.49      0.61      0.92           -2.30            1.31                0.10                3.69
salary     1.52      4.55      1.51           -1.44            4.47                0.24               87.63

              z    p  -log2(p)
seniority -0.54 0.59      0.76
salary     1.00 0.32      1.66
---
Concordance = 0.52
Log-likelihood ratio test = 1.37 on 2 df, -log2(p)=0.99
company 8 sales
********************
<lifelines.CoxPHFitter: fitted with 136 observations, 84 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 136
  number of events = 52
partial log-likelihood = -201.66
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.65      1.92      1.12           -1.55            2.85                0.21               17.37
salary    -2.00      0.14      2.51           -6.92            2.92                0.00               18.53

              z    p  -log2(p)
seniority  0.58 0.56      0.84
salary    -0.80 0.43      1.23
---
Concordance = 0.50
Log-likelihood ratio test = 0.69 on 2 df, -log2(p)=0.50
company 9 engineer
********************
<lifelines.CoxPHFitter: fitted with 185 observations, 104 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 185
  number of events = 81
partial log-likelihood = -355.88
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -1.25      0.29      0.77           -2.76            0.25                0.06                1.29
salary     0.92      2.52      1.27           -1.57            3.42                0.21               30.53

              z    p  -log2(p)
seniority -1.63 0.10      3.28
salary     0.72 0.47      1.09
---
Concordance = 0.55
Log-likelihood ratio test = 4.53 on 2 df, -log2(p)=3.27
company 9 marketing
********************
<lifelines.CoxPHFitter: fitted with 124 observations, 62 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 124
  number of events = 62
partial log-likelihood = -248.72
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.57      0.56      0.85           -2.24            1.10                0.11                3.01
salary     1.02      2.76      2.08           -3.06            5.09                0.05              162.65

              z    p  -log2(p)
seniority -0.67 0.50      0.99
salary     0.49 0.63      0.68
---
Concordance = 0.56
Log-likelihood ratio test = 0.48 on 2 df, -log2(p)=0.34
company 9 customer_service
********************
<lifelines.CoxPHFitter: fitted with 341 observations, 186 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 341
  number of events = 155
partial log-likelihood = -773.19
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.09      1.09      0.61           -1.11            1.29                0.33                3.63
salary     0.00      1.00      2.44           -4.78            4.78                0.01              119.59

             z    p  -log2(p)
seniority 0.15 0.88      0.18
salary    0.00 1.00      0.00
---
Concordance = 0.50
Log-likelihood ratio test = 0.10 on 2 df, -log2(p)=0.07
company 9 data_science
********************
<lifelines.CoxPHFitter: fitted with 133 observations, 70 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 133
  number of events = 63
partial log-likelihood = -253.93
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -0.07      0.93      0.84           -1.72            1.57                0.18                4.80
salary     0.89      2.43      1.27           -1.60            3.38                0.20               29.37

              z    p  -log2(p)
seniority -0.09 0.93      0.11
salary     0.70 0.48      1.04
---
Concordance = 0.57
Log-likelihood ratio test = 1.22 on 2 df, -log2(p)=0.88
company 9 sales
********************
<lifelines.CoxPHFitter: fitted with 113 observations, 63 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 113
  number of events = 50
partial log-likelihood = -192.57
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  1.18      3.26      1.00           -0.77            3.14                0.46               23.05
salary    -2.20      0.11      2.66           -7.41            3.02                0.00               20.50

              z    p  -log2(p)
seniority  1.18 0.24      2.08
salary    -0.83 0.41      1.29
---
Concordance = 0.52
Log-likelihood ratio test = 1.60 on 2 df, -log2(p)=1.16
company 10 engineer
********************
<lifelines.CoxPHFitter: fitted with 170 observations, 93 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 170
  number of events = 77
partial log-likelihood = -318.59
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.45      1.57      0.81           -1.14            2.03                0.32                7.64
salary    -1.97      0.14      1.42           -4.76            0.82                0.01                2.28

              z    p  -log2(p)
seniority  0.56 0.58      0.79
salary    -1.38 0.17      2.58
---
Concordance = 0.54
Log-likelihood ratio test = 3.39 on 2 df, -log2(p)=2.45
company 10 marketing
********************
<lifelines.CoxPHFitter: fitted with 96 observations, 56 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 96
  number of events = 40
partial log-likelihood = -151.41
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.25      1.29      1.00           -1.70            2.21                0.18                9.09
salary     0.50      1.65      2.53           -4.45            5.45                0.01              233.73

             z    p  -log2(p)
seniority 0.26 0.80      0.33
salary    0.20 0.84      0.25
---
Concordance = 0.58
Log-likelihood ratio test = 0.53 on 2 df, -log2(p)=0.39
company 10 customer_service
********************
<lifelines.CoxPHFitter: fitted with 333 observations, 189 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 333
  number of events = 144
partial log-likelihood = -711.79
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.97      2.65      0.60           -0.20            2.15                0.82                8.55
salary    -2.35      0.10      2.46           -7.17            2.47                0.00               11.87

              z    p  -log2(p)
seniority  1.63 0.10      3.27
salary    -0.95 0.34      1.56
---
Concordance = 0.54
Log-likelihood ratio test = 3.15 on 2 df, -log2(p)=2.27
company 10 data_science
********************
<lifelines.CoxPHFitter: fitted with 108 observations, 51 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 108
  number of events = 57
partial log-likelihood = -216.03
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority -2.80      0.06      1.03           -4.81           -0.78                0.01                0.46
salary     3.97     53.04      1.59            0.85            7.10                2.33             1206.08

              z    p  -log2(p)
seniority -2.72 0.01      7.27
salary     2.49 0.01      6.30
---
Concordance = 0.60
Log-likelihood ratio test = 7.73 on 2 df, -log2(p)=5.57
company 10 sales
********************
<lifelines.CoxPHFitter: fitted with 110 observations, 64 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 110
  number of events = 46
partial log-likelihood = -171.25
  time fit was run = 2019-07-19 01:24:38 UTC

---
           coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
seniority  0.98      2.68      1.03           -1.04            3.01                0.35               20.20
salary    -4.30      0.01      2.54           -9.29            0.68                0.00                1.98

              z    p  -log2(p)
seniority  0.95 0.34      1.56
salary    -1.69 0.09      3.46
---
Concordance = 0.63
Log-likelihood ratio test = 3.83 on 2 df, -log2(p)=2.77

# Using Cox Proportional Hazards model
for i in range(10):
    df_dummy2 = df_dummy[df_dummy.company_id == i+1]
    df_dummy2 = df_dummy2.drop(columns = ['company_id','seniority'])   
    
# Using Cox Proportional Hazards model
    for j,de in enumerate(depts):
        if j >1:
            data_test = df_dummy2[df_dummy2.dept == de]
            data_test = data_test.drop(columns = ['dept'])   
            cph = CoxPHFitter()   ## Instantiate the class to create a cph object
            cph.fit(data_test, 'lasting_days', event_col='event')   ## Fit the data to train the model
            print("company", i+1, de)
            print('*'*20)
            cph.print_summary()    ## HAve a look at the significance of the features
company 1 engineer
********************
<lifelines.CoxPHFitter: fitted with 1552 observations, 738 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1552
  number of events = 814
partial log-likelihood = -5206.53
  time fit was run = 2019-07-19 12:59:17 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.19      0.82      0.16           -0.52            0.13                0.60                1.14

           z    p  -log2(p)
salary -1.19 0.24      2.08
---
Concordance = 0.52
Log-likelihood ratio test = 1.39 on 1 df, -log2(p)=2.07
company 1 marketing
********************
<lifelines.CoxPHFitter: fitted with 1074 observations, 613 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1074
  number of events = 461
partial log-likelihood = -2791.89
  time fit was run = 2019-07-19 12:59:17 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.09      0.91      0.33           -0.74            0.56                0.48                1.74

           z    p  -log2(p)
salary -0.28 0.78      0.36
---
Concordance = 0.51
Log-likelihood ratio test = 0.08 on 1 df, -log2(p)=0.36
company 1 customer_service
********************
<lifelines.CoxPHFitter: fitted with 3129 observations, 1791 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 3129
  number of events = 1338
partial log-likelihood = -9540.69
  time fit was run = 2019-07-19 12:59:17 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.21      0.81      0.33           -0.84            0.43                0.43                1.54

           z    p  -log2(p)
salary -0.63 0.53      0.92
---
Concordance = 0.51
Log-likelihood ratio test = 0.40 on 1 df, -log2(p)=0.92
company 1 data_science
********************
<lifelines.CoxPHFitter: fitted with 1070 observations, 562 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1070
  number of events = 508
partial log-likelihood = -3089.99
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.41      1.51      0.22           -0.02            0.85                0.98                2.33

          z    p  -log2(p)
salary 1.86 0.06      4.00
---
Concordance = 0.53
Log-likelihood ratio test = 3.54 on 1 df, -log2(p)=4.06
company 1 sales
********************
<lifelines.CoxPHFitter: fitted with 1091 observations, 612 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1091
  number of events = 479
partial log-likelihood = -2916.82
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.04      1.04      0.33           -0.60            0.69                0.55                1.99

          z    p  -log2(p)
salary 0.13 0.90      0.16
---
Concordance = 0.51
Log-likelihood ratio test = 0.02 on 1 df, -log2(p)=0.16
company 2 engineer
********************
<lifelines.CoxPHFitter: fitted with 822 observations, 380 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 822
  number of events = 442
partial log-likelihood = -2599.59
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.09      1.10      0.23           -0.36            0.54                0.70                1.72

          z    p  -log2(p)
salary 0.40 0.69      0.53
---
Concordance = 0.51
Log-likelihood ratio test = 0.16 on 1 df, -log2(p)=0.53
company 2 marketing
********************
<lifelines.CoxPHFitter: fitted with 535 observations, 291 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 535
  number of events = 244
partial log-likelihood = -1324.51
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.27      0.76      0.46           -1.18            0.63                0.31                1.88

           z    p  -log2(p)
salary -0.59 0.55      0.85
---
Concordance = 0.52
Log-likelihood ratio test = 0.35 on 1 df, -log2(p)=0.85
company 2 customer_service
********************
<lifelines.CoxPHFitter: fitted with 1530 observations, 828 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1530
  number of events = 702
partial log-likelihood = -4468.09
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.25      0.78      0.46           -1.15            0.65                0.32                1.92

           z    p  -log2(p)
salary -0.54 0.59      0.76
---
Concordance = 0.51
Log-likelihood ratio test = 0.29 on 1 df, -log2(p)=0.76
company 2 data_science
********************
<lifelines.CoxPHFitter: fitted with 562 observations, 265 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 562
  number of events = 297
partial log-likelihood = -1617.68
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.08      0.92      0.30           -0.66            0.50                0.52                1.65

           z    p  -log2(p)
salary -0.27 0.79      0.34
---
Concordance = 0.50
Log-likelihood ratio test = 0.07 on 1 df, -log2(p)=0.34
company 2 sales
********************
<lifelines.CoxPHFitter: fitted with 508 observations, 291 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 508
  number of events = 217
partial log-likelihood = -1144.78
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.46      1.58      0.49           -0.50            1.41                0.61                4.11

          z    p  -log2(p)
salary 0.93 0.35      1.51
---
Concordance = 0.52
Log-likelihood ratio test = 0.88 on 1 df, -log2(p)=1.52
company 3 engineer
********************
<lifelines.CoxPHFitter: fitted with 512 observations, 291 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 512
  number of events = 221
partial log-likelihood = -1176.67
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.67      0.51      0.40           -1.46            0.12                0.23                1.13

           z    p  -log2(p)
salary -1.66 0.10      3.35
---
Concordance = 0.52
Log-likelihood ratio test = 2.68 on 1 df, -log2(p)=3.30
company 3 marketing
********************
<lifelines.CoxPHFitter: fitted with 367 observations, 213 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 367
  number of events = 154
partial log-likelihood = -771.21
  time fit was run = 2019-07-19 12:59:18 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.52      0.60      0.68           -1.85            0.82                0.16                2.26

           z    p  -log2(p)
salary -0.76 0.45      1.16
---
Concordance = 0.51
Log-likelihood ratio test = 0.58 on 1 df, -log2(p)=1.16
company 3 customer_service
********************
<lifelines.CoxPHFitter: fitted with 1000 observations, 538 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 1000
  number of events = 462
partial log-likelihood = -2799.56
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -1.30      0.27      0.68           -2.64            0.03                0.07                1.03

           z    p  -log2(p)
salary -1.92 0.06      4.18
---
Concordance = 0.51
Log-likelihood ratio test = 3.64 on 1 df, -log2(p)=4.15
company 3 data_science
********************
<lifelines.CoxPHFitter: fitted with 345 observations, 193 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 345
  number of events = 152
partial log-likelihood = -756.05
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.23      1.26      0.52           -0.78            1.25                0.46                3.48

          z    p  -log2(p)
salary 0.45 0.65      0.61
---
Concordance = 0.50
Log-likelihood ratio test = 0.20 on 1 df, -log2(p)=0.61
company 3 sales
********************
<lifelines.CoxPHFitter: fitted with 359 observations, 203 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 359
  number of events = 156
partial log-likelihood = -763.74
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  1.44      4.21      0.72            0.03            2.85                1.03               17.23

          z    p  -log2(p)
salary 2.00 0.05      4.45
---
Concordance = 0.56
Log-likelihood ratio test = 4.07 on 1 df, -log2(p)=4.52
company 4 engineer
********************
<lifelines.CoxPHFitter: fitted with 375 observations, 208 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 375
  number of events = 167
partial log-likelihood = -845.60
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.15      1.16      0.51           -0.85            1.15                0.43                3.17

          z    p  -log2(p)
salary 0.30 0.77      0.38
---
Concordance = 0.50
Log-likelihood ratio test = 0.09 on 1 df, -log2(p)=0.38
company 4 marketing
********************
<lifelines.CoxPHFitter: fitted with 263 observations, 153 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 263
  number of events = 110
partial log-likelihood = -514.29
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.42      0.66      0.88           -2.15            1.31                0.12                3.69

           z    p  -log2(p)
salary -0.48 0.63      0.66
---
Concordance = 0.52
Log-likelihood ratio test = 0.23 on 1 df, -log2(p)=0.66
company 4 customer_service
********************
<lifelines.CoxPHFitter: fitted with 769 observations, 412 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 769
  number of events = 357
partial log-likelihood = -2051.02
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.67      1.95      0.81           -0.92            2.26                0.40                9.57

          z    p  -log2(p)
salary 0.83 0.41      1.29
---
Concordance = 0.51
Log-likelihood ratio test = 0.69 on 1 df, -log2(p)=1.30
company 4 data_science
********************
<lifelines.CoxPHFitter: fitted with 277 observations, 159 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 277
  number of events = 118
partial log-likelihood = -546.81
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.26      1.30      0.61           -0.94            1.46                0.39                4.31

          z    p  -log2(p)
salary 0.42 0.67      0.57
---
Concordance = 0.52
Log-likelihood ratio test = 0.18 on 1 df, -log2(p)=0.58
company 4 sales
********************
<lifelines.CoxPHFitter: fitted with 252 observations, 150 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 252
  number of events = 102
partial log-likelihood = -468.70
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.40      1.49      0.93           -1.42            2.23                0.24                9.28

          z    p  -log2(p)
salary 0.43 0.67      0.59
---
Concordance = 0.51
Log-likelihood ratio test = 0.19 on 1 df, -log2(p)=0.59
company 5 engineer
********************
<lifelines.CoxPHFitter: fitted with 311 observations, 178 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 311
  number of events = 133
partial log-likelihood = -647.13
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -1.22      0.30      0.55           -2.30           -0.13                0.10                0.87

           z    p  -log2(p)
salary -2.20 0.03      5.18
---
Concordance = 0.52
Log-likelihood ratio test = 4.71 on 1 df, -log2(p)=5.06
company 5 marketing
********************
<lifelines.CoxPHFitter: fitted with 224 observations, 114 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 224
  number of events = 110
partial log-likelihood = -497.49
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.05      1.05      0.81           -1.53            1.63                0.22                5.12

          z    p  -log2(p)
salary 0.06 0.95      0.07
---
Concordance = 0.51
Log-likelihood ratio test = 0.00 on 1 df, -log2(p)=0.07
company 5 customer_service
********************
<lifelines.CoxPHFitter: fitted with 631 observations, 353 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 631
  number of events = 278
partial log-likelihood = -1545.97
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.91      0.40      0.90           -2.68            0.86                0.07                2.37

           z    p  -log2(p)
salary -1.01 0.31      1.67
---
Concordance = 0.51
Log-likelihood ratio test = 1.00 on 1 df, -log2(p)=1.66
company 5 data_science
********************
<lifelines.CoxPHFitter: fitted with 213 observations, 113 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 213
  number of events = 100
partial log-likelihood = -440.81
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.50      0.61      0.66           -1.80            0.81                0.17                2.24

           z    p  -log2(p)
salary -0.75 0.46      1.13
---
Concordance = 0.52
Log-likelihood ratio test = 0.55 on 1 df, -log2(p)=1.12
company 5 sales
********************
<lifelines.CoxPHFitter: fitted with 254 observations, 148 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 254
  number of events = 106
partial log-likelihood = -483.49
  time fit was run = 2019-07-19 12:59:19 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.12      0.89      0.88           -1.83            1.60                0.16                4.94

           z    p  -log2(p)
salary -0.14 0.89      0.16
---
Concordance = 0.48
Log-likelihood ratio test = 0.02 on 1 df, -log2(p)=0.16
company 6 engineer
********************
<lifelines.CoxPHFitter: fitted with 218 observations, 115 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 218
  number of events = 103
partial log-likelihood = -453.19
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.05      1.05      0.60           -1.12            1.22                0.33                3.40

          z    p  -log2(p)
salary 0.09 0.93      0.10
---
Concordance = 0.51
Log-likelihood ratio test = 0.01 on 1 df, -log2(p)=0.10
company 6 marketing
********************
<lifelines.CoxPHFitter: fitted with 174 observations, 117 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 174
  number of events = 57
partial log-likelihood = -252.45
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.20      1.22      1.17           -2.10            2.50                0.12               12.19

          z    p  -log2(p)
salary 0.17 0.86      0.21
---
Concordance = 0.50
Log-likelihood ratio test = 0.03 on 1 df, -log2(p)=0.21
company 6 customer_service
********************
<lifelines.CoxPHFitter: fitted with 495 observations, 257 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 495
  number of events = 238
partial log-likelihood = -1260.22
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.71      2.04      0.98           -1.20            2.63                0.30               13.86

          z    p  -log2(p)
salary 0.73 0.46      1.10
---
Concordance = 0.53
Log-likelihood ratio test = 0.54 on 1 df, -log2(p)=1.11
company 6 data_science
********************
<lifelines.CoxPHFitter: fitted with 151 observations, 84 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 151
  number of events = 67
partial log-likelihood = -272.89
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -1.13      0.32      0.76           -2.61            0.35                0.07                1.42

           z    p  -log2(p)
salary -1.50 0.13      2.90
---
Concordance = 0.57
Log-likelihood ratio test = 2.18 on 1 df, -log2(p)=2.84
company 6 sales
********************
<lifelines.CoxPHFitter: fitted with 161 observations, 85 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 161
  number of events = 76
partial log-likelihood = -320.96
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.22      1.24      1.06           -1.85            2.29                0.16                9.85

          z    p  -log2(p)
salary 0.21 0.84      0.26
---
Concordance = 0.49
Log-likelihood ratio test = 0.04 on 1 df, -log2(p)=0.26
company 7 engineer
********************
<lifelines.CoxPHFitter: fitted with 223 observations, 123 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 223
  number of events = 100
partial log-likelihood = -445.48
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.59      1.80      0.62           -0.63            1.81                0.53                6.09

          z    p  -log2(p)
salary 0.94 0.35      1.53
---
Concordance = 0.54
Log-likelihood ratio test = 0.90 on 1 df, -log2(p)=1.55
company 7 marketing
********************
<lifelines.CoxPHFitter: fitted with 140 observations, 77 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 140
  number of events = 63
partial log-likelihood = -244.08
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.09      1.09      1.13           -2.12            2.29                0.12                9.92

          z    p  -log2(p)
salary 0.08 0.94      0.09
---
Concordance = 0.57
Log-likelihood ratio test = 0.01 on 1 df, -log2(p)=0.09
company 7 customer_service
********************
<lifelines.CoxPHFitter: fitted with 466 observations, 266 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 466
  number of events = 200
partial log-likelihood = -1045.27
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.42      1.52      1.09           -1.72            2.57                0.18               13.00

          z    p  -log2(p)
salary 0.39 0.70      0.51
---
Concordance = 0.51
Log-likelihood ratio test = 0.15 on 1 df, -log2(p)=0.52
company 7 data_science
********************
<lifelines.CoxPHFitter: fitted with 149 observations, 84 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 149
  number of events = 65
partial log-likelihood = -270.28
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.71      0.49      0.73           -2.14            0.72                0.12                2.05

           z    p  -log2(p)
salary -0.98 0.33      1.60
---
Concordance = 0.52
Log-likelihood ratio test = 0.94 on 1 df, -log2(p)=1.59
company 7 sales
********************
<lifelines.CoxPHFitter: fitted with 162 observations, 96 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 162
  number of events = 66
partial log-likelihood = -270.22
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.57      0.57      1.07           -2.66            1.52                0.07                4.57

           z    p  -log2(p)
salary -0.53 0.59      0.75
---
Concordance = 0.50
Log-likelihood ratio test = 0.28 on 1 df, -log2(p)=0.75
company 8 engineer
********************
<lifelines.CoxPHFitter: fitted with 190 observations, 102 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 190
  number of events = 88
partial log-likelihood = -380.02
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  1.07      2.90      0.78           -0.47            2.60                0.63               13.48

          z    p  -log2(p)
salary 1.36 0.17      2.53
---
Concordance = 0.53
Log-likelihood ratio test = 1.92 on 1 df, -log2(p)=2.59
company 8 marketing
********************
<lifelines.CoxPHFitter: fitted with 132 observations, 67 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 132
  number of events = 65
partial log-likelihood = -249.54
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  2.23      9.28      1.16           -0.04            4.50                0.96               90.02

          z    p  -log2(p)
salary 1.92 0.05      4.19
---
Concordance = 0.55
Log-likelihood ratio test = 3.74 on 1 df, -log2(p)=4.24
company 8 customer_service
********************
<lifelines.CoxPHFitter: fitted with 378 observations, 213 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 378
  number of events = 165
partial log-likelihood = -833.59
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.68      1.96      1.20           -1.67            3.02                0.19               20.52

          z    p  -log2(p)
salary 0.56 0.57      0.80
---
Concordance = 0.51
Log-likelihood ratio test = 0.32 on 1 df, -log2(p)=0.81
company 8 data_science
********************
<lifelines.CoxPHFitter: fitted with 143 observations, 78 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 143
  number of events = 65
partial log-likelihood = -261.90
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.82      2.28      0.80           -0.75            2.40                0.47               11.01

          z    p  -log2(p)
salary 1.02 0.31      1.70
---
Concordance = 0.53
Log-likelihood ratio test = 1.08 on 1 df, -log2(p)=1.74
company 8 sales
********************
<lifelines.CoxPHFitter: fitted with 136 observations, 84 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 136
  number of events = 52
partial log-likelihood = -201.83
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.74      0.48      1.25           -3.19            1.70                0.04                5.50

           z    p  -log2(p)
salary -0.60 0.55      0.86
---
Concordance = 0.49
Log-likelihood ratio test = 0.36 on 1 df, -log2(p)=0.86
company 9 engineer
********************
<lifelines.CoxPHFitter: fitted with 185 observations, 104 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 185
  number of events = 81
partial log-likelihood = -357.23
  time fit was run = 2019-07-19 12:59:20 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.89      0.41      0.65           -2.18            0.39                0.11                1.48

           z    p  -log2(p)
salary -1.37 0.17      2.54
---
Concordance = 0.51
Log-likelihood ratio test = 1.83 on 1 df, -log2(p)=2.50
company 9 marketing
********************
<lifelines.CoxPHFitter: fitted with 124 observations, 62 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 124
  number of events = 62
partial log-likelihood = -248.95
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.18      0.84      1.11           -2.35            2.00                0.10                7.35

           z    p  -log2(p)
salary -0.16 0.87      0.19
---
Concordance = 0.50
Log-likelihood ratio test = 0.03 on 1 df, -log2(p)=0.19
company 9 customer_service
********************
<lifelines.CoxPHFitter: fitted with 341 observations, 186 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 341
  number of events = 155
partial log-likelihood = -773.20
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.32      1.37      1.14           -1.91            2.54                0.15               12.72

          z    p  -log2(p)
salary 0.28 0.78      0.36
---
Concordance = 0.50
Log-likelihood ratio test = 0.08 on 1 df, -log2(p)=0.36
company 9 data_science
********************
<lifelines.CoxPHFitter: fitted with 133 observations, 70 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 133
  number of events = 63
partial log-likelihood = -253.94
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.80      2.22      0.73           -0.64            2.23                0.53                9.33

          z    p  -log2(p)
salary 1.09 0.28      1.85
---
Concordance = 0.57
Log-likelihood ratio test = 1.21 on 1 df, -log2(p)=1.88
company 9 sales
********************
<lifelines.CoxPHFitter: fitted with 113 observations, 63 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 113
  number of events = 50
partial log-likelihood = -193.26
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.57      1.76      1.22           -1.83            2.97                0.16               19.41

          z    p  -log2(p)
salary 0.46 0.64      0.64
---
Concordance = 0.48
Log-likelihood ratio test = 0.22 on 1 df, -log2(p)=0.64
company 10 engineer
********************
<lifelines.CoxPHFitter: fitted with 170 observations, 93 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 170
  number of events = 77
partial log-likelihood = -318.74
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -1.29      0.28      0.72           -2.71            0.13                0.07                1.13

           z    p  -log2(p)
salary -1.79 0.07      3.75
---
Concordance = 0.55
Log-likelihood ratio test = 3.08 on 1 df, -log2(p)=3.66
company 10 marketing
********************
<lifelines.CoxPHFitter: fitted with 96 observations, 56 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 96
  number of events = 40
partial log-likelihood = -151.44
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  1.02      2.77      1.50           -1.92            3.96                0.15               52.37

          z    p  -log2(p)
salary 0.68 0.50      1.01
---
Concordance = 0.58
Log-likelihood ratio test = 0.47 on 1 df, -log2(p)=1.02
company 10 customer_service
********************
<lifelines.CoxPHFitter: fitted with 333 observations, 189 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 333
  number of events = 144
partial log-likelihood = -713.10
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary  0.97      2.64      1.33           -1.64            3.58                0.19               35.85

          z    p  -log2(p)
salary 0.73 0.47      1.10
---
Concordance = 0.54
Log-likelihood ratio test = 0.54 on 1 df, -log2(p)=1.11
company 10 data_science
********************
<lifelines.CoxPHFitter: fitted with 108 observations, 51 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 108
  number of events = 57
partial log-likelihood = -219.88
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -0.09      0.91      0.71           -1.48            1.29                0.23                3.64

           z    p  -log2(p)
salary -0.13 0.89      0.16
---
Concordance = 0.50
Log-likelihood ratio test = 0.02 on 1 df, -log2(p)=0.16
company 10 sales
********************
<lifelines.CoxPHFitter: fitted with 110 observations, 64 censored>
      duration col = 'lasting_days'
         event col = 'event'
number of subjects = 110
  number of events = 46
partial log-likelihood = -171.71
  time fit was run = 2019-07-19 12:59:21 UTC

---
        coef exp(coef)  se(coef)  coef lower 95%  coef upper 95% exp(coef) lower 95% exp(coef) upper 95%
salary -2.27      0.10      1.31           -4.84            0.30                0.01                1.35

           z    p  -log2(p)
salary -1.73 0.08      3.58
---
Concordance = 0.64
Log-likelihood ratio test = 2.93 on 1 df, -log2(p)=3.52

About

A case study to estimate employee turn over rate using machine learning

Resources

Readme
Activity

Stars

2 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages