mod_7_console_1.txt

Python 3.9.13 (tags/v3.9.13:6de2ca5, May 17 2022, 16:36:42) [MSC v.1929 64 bit (AMD64)]
Type "copyright", "credits" or "license" for more information.

IPython  -- An enhanced Interactive Python.

In [1]: import numpy as np   # import numpy into Python and use it under its alias 'np'
   ...: import pandas as pd  # import pandas into Python and use it under its alias 'pd'
   ...: import statsmodels.api as sm  # import statsmodels.api into Python and use it under its alias 'sm'
   ...: import sklearn # import sklearn into Python 
   ...: import matplotlib.pyplot as plt # import matplotlib.pyplot into Python and use it under its alias 'plt'
   ...: import seaborn as sns # import the seaborn library and use it under its alias 'sns'
   ...: from scipy import stats,integrate 
   ...: import pylab # import pylab into Python
   ...: import scipy.stats as stats  # import scipy.stats into Python and use it under its alias 'stats'
   ...: 
   ...: # Now we can read/import the data into Python. The data is included in a typical .csv data file.

In [2]: Telecom_Churn_Data = pd.read_csv("C:/Users/rao8r/Downloads/mod-7/to-sandhya/TelecomCustomerChurnData.csv", sep =",")

In [3]: Telecom_Churn_Data.head(20) # show the first 20 rows in the data
Out[3]: 
    Tenure PhoneService  ... TotalCharges Churn
0        1           No  ...        29.85     0
1       34          Yes  ...      1889.50     0
2        2          Yes  ...       108.15     1
3       45           No  ...      1840.75     0
4        2          Yes  ...       151.65     1
5        8          Yes  ...       820.50     1
6       22          Yes  ...      1949.40     0
7       10           No  ...       301.90     0
8       28          Yes  ...      3046.05     1
9       62          Yes  ...      3487.95     0
10      13          Yes  ...       587.45     0
11      16          Yes  ...       326.80     0
12      58          Yes  ...      5681.10     0
13      49          Yes  ...      5036.30     1
14      25          Yes  ...      2686.05     0
15      69          Yes  ...      7895.15     0
16      52          Yes  ...      1022.95     0
17      71          Yes  ...      7382.25     0
18      10          Yes  ...       528.35     1
19      21          Yes  ...      1862.90     0

[20 rows x 7 columns]

In [4]: Telecom_Churn_Data.tail(20) # show the last 20 rows in the data
Out[4]: 
      Tenure PhoneService  ... TotalCharges Churn
7011      72          Yes  ...      7544.30     0
7012      63          Yes  ...      6479.40     0
7013      44          Yes  ...      3626.35     0
7014      18          Yes  ...      1679.40     0
7015       9          Yes  ...       403.35     1
7016      13          Yes  ...       931.55     0
7017      68          Yes  ...      4326.25     0
7018       6           No  ...       263.05     0
7019       2          Yes  ...        39.25     0
7020      55          Yes  ...      3316.10     0
7021       1          Yes  ...        75.75     1
7022      38          Yes  ...      2625.25     0
7023      67          Yes  ...      6886.25     1
7024      19          Yes  ...      1495.10     0
7025      12           No  ...       743.30     0
7026      72          Yes  ...      1419.40     0
7027      24          Yes  ...      1990.50     0
7028      72          Yes  ...      7362.90     0
7029      11           No  ...       346.45     0
7030       4          Yes  ...       306.60     1

[20 rows x 7 columns]

In [5]: Telecom_Churn_Data.columns.tolist() # show the names of columns/variables in the data
Out[5]: 
['Tenure',
 'PhoneService',
 'Contract',
 'PaperlessBilling',
 'PaymentMethod',
 'TotalCharges',
 'Churn']

In [6]: Telecom_Churn_Data.shape # output the dimension of the Telecom_Churn_Data object. This is similar to dim() in R
Out[6]: (7031, 7)

In [7]: Telecom_Churn_Data.dtypes # show the data types of the variables in the data
Out[7]: 
Tenure                int64
PhoneService         object
Contract             object
PaperlessBilling     object
PaymentMethod        object
TotalCharges        float64
Churn                 int64
dtype: object

In [8]: Stat_summary_table = Telecom_Churn_Data.describe().T # output the statistics for all the variables.

In [9]: check_missing_value = Telecom_Churn_Data.isnull().sum(axis=1) 

In [10]: check_missing_value[check_missing_value!=0].count() 
Out[10]: 0

In [11]: Telecom_Churn_Data.info() # you can check whther there are "null's" in each variable.
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7031 entries, 0 to 7030
Data columns (total 7 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   Tenure            7031 non-null   int64  
 1   PhoneService      7031 non-null   object 
 2   Contract          7031 non-null   object 
 3   PaperlessBilling  7031 non-null   object 
 4   PaymentMethod     7031 non-null   object 
 5   TotalCharges      7031 non-null   float64
 6   Churn             7031 non-null   int64  
dtypes: float64(1), int64(2), object(4)
memory usage: 384.6+ KB

In [12]: List_Cate_Var = ["Tenure","PhoneService", "Contract", "PaperlessBilling", "PaymentMethod", "TotalCharges", "Churn"] 

In [13]: for var_name in List_Cate_Var:
    ...:     print("The Frequency Table of the", var_name, "Variable")
    ...:     print(Telecom_Churn_Data[var_name].value_counts()) # generate the frequency table for each categorical variable in the loop
The Frequency Table of the Tenure Variable
1     613
72    362
2     238
3     200
4     176
     ... 
38     59
28     57
39     56
44     51
36     50
Name: Tenure, Length: 72, dtype: int64
The Frequency Table of the PhoneService Variable
Yes    6351
No      680
Name: PhoneService, dtype: int64
The Frequency Table of the Contract Variable
Month-to-month    3875
Two year          1684
One year          1472
Name: Contract, dtype: int64
The Frequency Table of the PaperlessBilling Variable
Yes    4167
No     2864
Name: PaperlessBilling, dtype: int64
The Frequency Table of the PaymentMethod Variable
Electronic check             2365
Mailed check                 1604
Bank transfer (automatic)    1541
Credit card (automatic)      1521
Name: PaymentMethod, dtype: int64
The Frequency Table of the TotalCharges Variable
20.20      11
19.75       9
19.90       8
20.05       8
19.65       8
           ..
6849.40     1
692.35      1
130.15      1
3211.90     1
306.60      1
Name: TotalCharges, Length: 6529, dtype: int64
The Frequency Table of the Churn Variable
0    5162
1    1869
Name: Churn, dtype: int64

In [14]: plt.figure() # open a new figure window
Out[14]: <Figure size 432x288 with 0 Axes><Figure size 432x288 with 0 Axes>

In [15]: plt.hist(Telecom_Churn_Data['Tenure'], bins=25) # generate a histogram with the number of bins = 25
Out[15]: 
(array([1051.,  419.,  373.,  332.,  284.,  264.,  144.,  238.,  252.,
         201.,  206.,  217.,  115.,  179.,  200.,  186.,  198.,  216.,
         138.,  209.,  203.,  218.,  244.,  293.,  651.]),
 array([ 1.  ,  3.84,  6.68,  9.52, 12.36, 15.2 , 18.04, 20.88, 23.72,
        26.56, 29.4 , 32.24, 35.08, 37.92, 40.76, 43.6 , 46.44, 49.28,
        52.12, 54.96, 57.8 , 60.64, 63.48, 66.32, 69.16, 72.  ]),
 <BarContainer object of 25 artists>)
 
Warning
Figures now render in the Plots pane by default. To make them also appear inline in the Console, uncheck "Mute Inline Plotting" under the Plots pane options menu.
 

In [16]: plt.hist(Telecom_Churn_Data['Tenure'], bins=50) # generate a histogram with the number of bins = 25
Out[16]: 
(array([851., 200., 309., 110., 254., 119., 116., 216., 109., 175.,  80.,
        184.,  73.,  71., 153.,  85., 173.,  79.,  72., 129.,  72., 134.,
         64., 153.,  50.,  65., 115.,  64., 135.,  65., 112.,  74.,  68.,
        130.,  68., 148.,  70.,  68., 144.,  65., 127.,  76., 146.,  72.,
         80., 164.,  98., 195., 119., 532.]),
 array([ 1.  ,  2.42,  3.84,  5.26,  6.68,  8.1 ,  9.52, 10.94, 12.36,
        13.78, 15.2 , 16.62, 18.04, 19.46, 20.88, 22.3 , 23.72, 25.14,
        26.56, 27.98, 29.4 , 30.82, 32.24, 33.66, 35.08, 36.5 , 37.92,
        39.34, 40.76, 42.18, 43.6 , 45.02, 46.44, 47.86, 49.28, 50.7 ,
        52.12, 53.54, 54.96, 56.38, 57.8 , 59.22, 60.64, 62.06, 63.48,
        64.9 , 66.32, 67.74, 69.16, 70.58, 72.  ]),
 <BarContainer object of 50 artists>)

In [17]: plt.hist(Telecom_Churn_Data['Tenure'], bins=10) # generate a histogram with the number of bins = 25
Out[17]: 
(array([1724.,  735.,  561.,  538.,  473.,  444.,  452.,  495.,  501.,
        1108.]),
 array([ 1. ,  8.1, 15.2, 22.3, 29.4, 36.5, 43.6, 50.7, 57.8, 64.9, 72. ]),
 <BarContainer object of 10 artists>)

In [18]: plt.xlabel("Tenure")  # add a x-label
Out[18]: Text(0.5, 0, 'Tenure')

In [19]: plt.ylabel("Frequency") # add a y-label
Out[19]: Text(0, 0.5, 'Frequency')

In [20]: plt.title("Histogram of Tenure") # add a title for the chart
Out[20]: Text(0.5, 1.0, 'Histogram of Tenure')

In [21]: plt.figure() # open a new figure window
    ...: plt.xlabel("Tenure")  # add a x-label
    ...: plt.ylabel("Frequency") # add a y-label
    ...: plt.title("Histogram of Tenure") # add a title for the chart
    ...: plt.hist(Telecom_Churn_Data['Tenure'], bins=25) # generate a histogram with the number of bins = 25
    ...: # here bins = 10 means that 25 classes/groups are specified
Out[21]: 
(array([1051.,  419.,  373.,  332.,  284.,  264.,  144.,  238.,  252.,
         201.,  206.,  217.,  115.,  179.,  200.,  186.,  198.,  216.,
         138.,  209.,  203.,  218.,  244.,  293.,  651.]),
 array([ 1.  ,  3.84,  6.68,  9.52, 12.36, 15.2 , 18.04, 20.88, 23.72,
        26.56, 29.4 , 32.24, 35.08, 37.92, 40.76, 43.6 , 46.44, 49.28,
        52.12, 54.96, 57.8 , 60.64, 63.48, 66.32, 69.16, 72.  ]),
 <BarContainer object of 25 artists>)

In [22]: plt.figure() # open a new figure window
    ...: sns.kdeplot(Telecom_Churn_Data['Tenure']) # 'kde' here means a kernel density
Out[22]: <AxesSubplot:xlabel='Tenure', ylabel='Density'>

In [23]: plt.figure() # open a new figure window
    ...: sns.distplot(Telecom_Churn_Data['Tenure']) # as you can see that sns.distplot automatically choose
C:\Users\rao8r\AppData\Local\Programs\Python\Python39\lib\site-packages\seaborn\distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
Out[23]: <AxesSubplot:xlabel='Tenure', ylabel='Density'>

In [24]: plt.figure() # open a new figure window                                                             
    ...: sns.boxplot(Telecom_Churn_Data['Tenure'])
C:\Users\rao8r\AppData\Local\Programs\Python\Python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
Out[24]: <AxesSubplot:xlabel='Tenure'>

In [25]: 
    ...: plt.figure() # open a new figure window
    ...: plt.xlabel("TotalCharges")  # add a x-label
    ...: plt.ylabel("Frequency") # add a y-label
    ...: plt.title("Histogram of TotalCharges") # add a title for the chart
    ...: plt.hist(Telecom_Churn_Data['TotalCharges'], bins=25) # generate a histogram with the number of bins = 25
    ...: # here bins = 10 means that 25 classes/groups are specified
Out[25]: 
(array([1678.,  728.,  587.,  536.,  408.,  318.,  243.,  215.,  208.,
         195.,  173.,  196.,  165.,  174.,  158.,  146.,  164.,  148.,
         135.,  111.,  101.,   87.,   77.,   55.,   25.]),
 array([  18.8 ,  365.44,  712.08, 1058.72, 1405.36, 1752.  , 2098.64,
        2445.28, 2791.92, 3138.56, 3485.2 , 3831.84, 4178.48, 4525.12,
        4871.76, 5218.4 , 5565.04, 5911.68, 6258.32, 6604.96, 6951.6 ,
        7298.24, 7644.88, 7991.52, 8338.16, 8684.8 ]),
 <BarContainer object of 25 artists>)

In [26]: 
    ...: plt.figure() # open a new figure window
    ...: plt.xlabel("TotalCharges")  # add a x-label
    ...: plt.ylabel("Frequency") # add a y-label
    ...: plt.title("Histogram of TotalCharges") # add a title for the chart
    ...: plt.hist(Telecom_Churn_Data['TotalCharges'], bins=15) # generate a histogram with the number of bins = 25
    ...: # here bins = 10 means that 25 classes/groups are specified
Out[26]: 
(array([2197.,  970.,  770.,  477.,  364.,  338.,  309.,  281.,  276.,
         267.,  236.,  201.,  157.,  130.,   58.]),
 array([  18.8       ,  596.53333333, 1174.26666667, 1752.        ,
        2329.73333333, 2907.46666667, 3485.2       , 4062.93333333,
        4640.66666667, 5218.4       , 5796.13333333, 6373.86666667,
        6951.6       , 7529.33333333, 8107.06666667, 8684.8       ]),
 <BarContainer object of 15 artists>)

In [27]: plt.figure() # open a new figure window
    ...: sns.kdeplot(Telecom_Churn_Data['TotalCharges']) # 'kde' here means a kernel density
Out[27]: <AxesSubplot:xlabel='TotalCharges', ylabel='Density'>

In [28]: 
    ...: plt.figure() # open a new figure window
    ...: sns.distplot(Telecom_Churn_Data['TotalCharges']) # as you can see that sns.distplot automatically choose
C:\Users\rao8r\AppData\Local\Programs\Python\Python39\lib\site-packages\seaborn\distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)
Out[28]: <AxesSubplot:xlabel='TotalCharges', ylabel='Density'>

In [29]: 
    ...: 
    ...: plt.figure() # open a new figure window                                                             
    ...: sns.boxplot(Telecom_Churn_Data['TotalCharges'])
C:\Users\rao8r\AppData\Local\Programs\Python\Python39\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
Out[29]: <AxesSubplot:xlabel='TotalCharges'>

In [30]: from scipy.stats import chi2_contingency # import the chi2_contingency module from the scipy.stats

In [31]: chi2_contingency(pd.crosstab(Telecom_Churn_Data['Churn'], Telecom_Churn_Data['Tenure']))
Out[31]: 
(1059.9449494025498,
 4.084542332671357e-176,
 71,
 array([[450.05063291, 174.73417722, 146.83544304, 129.21518987,
          97.64556962,  80.75949367,  96.17721519,  90.30379747,
          87.36708861,  85.16455696,  72.6835443 ,  85.89873418,
          80.02531646,  55.79746835,  72.6835443 ,  58.73417722,
          63.87341772,  71.21518987,  53.59493671,  52.12658228,
          46.25316456,  66.07594937,  62.40506329,  69.01265823,
          58.        ,  58.        ,  52.86075949,  41.84810127,
          52.86075949,  52.86075949,  47.72151899,  50.65822785,
          46.98734177,  47.72151899,  64.60759494,  36.70886076,
          47.72151899,  43.3164557 ,  41.11392405,  46.98734177,
          51.39240506,  47.72151899,  47.72151899,  37.44303797,
          44.78481013,  54.32911392,  49.92405063,  46.98734177,
          48.4556962 ,  49.92405063,  49.92405063,  58.73417722,
          51.39240506,  49.92405063,  46.98734177,  58.73417722,
          47.72151899,  49.18987342,  44.05063291,  55.79746835,
          55.79746835,  51.39240506,  52.86075949,  58.73417722,
          55.79746835,  64.60759494,  71.94936709,  73.41772152,
          69.74683544,  87.36708861, 124.81012658, 265.7721519 ],
        [162.94936709,  63.26582278,  53.16455696,  46.78481013,
          35.35443038,  29.24050633,  34.82278481,  32.69620253,
          31.63291139,  30.83544304,  26.3164557 ,  31.10126582,
          28.97468354,  20.20253165,  26.3164557 ,  21.26582278,
          23.12658228,  25.78481013,  19.40506329,  18.87341772,
          16.74683544,  23.92405063,  22.59493671,  24.98734177,
          21.        ,  21.        ,  19.13924051,  15.15189873,
          19.13924051,  19.13924051,  17.27848101,  18.34177215,
          17.01265823,  17.27848101,  23.39240506,  13.29113924,
          17.27848101,  15.6835443 ,  14.88607595,  17.01265823,
          18.60759494,  17.27848101,  17.27848101,  13.55696203,
          16.21518987,  19.67088608,  18.07594937,  17.01265823,
          17.5443038 ,  18.07594937,  18.07594937,  21.26582278,
          18.60759494,  18.07594937,  17.01265823,  21.26582278,
          17.27848101,  17.81012658,  15.94936709,  20.20253165,
          20.20253165,  18.60759494,  19.13924051,  21.26582278,
          20.20253165,  23.39240506,  26.05063291,  26.58227848,
          25.25316456,  31.63291139,  45.18987342,  96.2278481 ]]))

In [32]: chi2_contingency(pd.crosstab(Telecom_Churn_Data['Churn'], Telecom_Churn_Data['PhoneService']))
Out[32]: 
(0.8780612721866795,
 0.3487332316441293,
 1,
 array([[ 499.24050633, 4662.75949367],
        [ 180.75949367, 1688.24050633]]))

In [33]: chi2_contingency(pd.crosstab(Telecom_Churn_Data['Churn'], Telecom_Churn_Data['Contract']))
Out[33]: 
(1179.080956754422,
 9.243221963845809e-257,
 2,
 array([[2844.93670886, 1080.70886076, 1236.35443038],
        [1030.06329114,  391.29113924,  447.64556962]]))

In [34]: chi2_contingency(pd.crosstab(Telecom_Churn_Data['Churn'], Telecom_Churn_Data['PaperlessBilling']))
Out[34]: 
(257.06741761837947,
 7.477609371069561e-58,
 1,
 array([[2102.6835443, 3059.3164557],
        [ 761.3164557, 1107.6835443]]))

In [35]: chi2_contingency(pd.crosstab(Telecom_Churn_Data['Churn'], Telecom_Churn_Data['PaymentMethod']))
Out[35]: 
(645.1528611114002,
 1.6378691391382898e-139,
 3,
 array([[1131.36708861, 1116.6835443 , 1736.32911392, 1177.62025316],
        [ 409.63291139,  404.3164557 ,  628.67088608,  426.37974684]]))

In [36]: chi2_contingency(pd.crosstab(Telecom_Churn_Data['Churn'], Telecom_Churn_Data['TotalCharges']))
Out[36]: 
(6502.6232970687415,
 0.5857175785198268,
 6528,
 array([[0.73417722, 1.46835443, 0.73417722, ..., 0.73417722, 0.73417722,
         0.73417722],
        [0.26582278, 0.53164557, 0.26582278, ..., 0.26582278, 0.26582278,
         0.26582278]]))

371069561e-58,
 1,
 array([[2102.6835443, 3059.3164557],
        [ 761.3164557, 1107.6835443]]))

In [35]: chi2_contingency(pd.crosstab(Telecom_Churn_Data['Churn'], Telecom_Churn_Data['PaymentMethod']))
Out[35]: 
(645.1528611114002,
 1.6378691391382898e-139,
 3,
 array([[1131.36708861, 1116.6835443 , 1736.32911392, 1177.62025316],
        [ 409.63291139,  404.3164557 ,  628.67088608,  426.37974684]]))

In [36]: chi2_contingency(pd.crosstab(Telecom_Churn_Data['Churn'], Telecom_Churn_Data['TotalCharges']))
Out[36]: 
(6502.6232970687415,
 0.5857175785198268,
 6528,
 array([[0.73417722, 1.46835443, 0.73417722, ..., 0.73417722, 0.73417722,
         0.73417722],
        [0.26582278, 0.53164557, 0.26582278, ..., 0.26582278, 0.26582278,
         0.26582278]]))

In [37]: Telecom_Churn_Data['PhoneService'] = Telecom_Churn_Data['PhoneService'].astype('category')

In [38]: Telecom_Churn_Data['Contract'] = Telecom_Churn_Data['Contract'].astype('category')

In [39]: Telecom_Churn_Data['PaperlessBilling'] = Telecom_Churn_Data['PaperlessBilling'].astype('category')

In [40]: Telecom_Churn_Data['PaymentMethod'] = Telecom_Churn_Data['PaymentMethod'].astype('category')

In [41]: indep_column_list = ['Tenure', 'Contract', 'PaperlessBilling', 'PaymentMethod', 'TotalCharges']

In [42]: indep_column_list # take a look at the new list. 
Out[42]: ['Tenure', 'Contract', 'PaperlessBilling', 'PaymentMethod', 'TotalCharges']

In [43]: # Now we can create a dataframe with dummies being created for all the categorical independent variables,

In [44]: # for the numeric variables such as Age, it remains to be numeric in the dataframe:

In [45]: indep_dummy_data = pd.get_dummies(Telecom_Churn_Data[indep_column_list], drop_first = True)

In [46]: # note that drop_first = True means that the first level in each categorical variable is skipped when creating the dummies.

In [47]: indep_dummy_data.head()  # take a look at the independent variables' dataframe after the dummies are created. The first five rows.
Out[47]: 
   Tenure  ...  PaymentMethod_Mailed check
0       1  ...                           0
1      34  ...                           1
2       2  ...                           1
3      45  ...                           0
4       2  ...                           0

[5 rows x 8 columns]

In [48]: len(indep_dummy_data.columns)  # output the total number of the columns in the dataframe
Out[48]: 8

In [49]: # Now let's run the logistic regression modeling 

In [50]: # We can use the statsmodels.api's sm.Logit to fit a logit model

In [51]: # before fitting the logit model, make sure to include intercept in the model by adding a new column to the 

In [52]: # independent variables' dataframe:

In [53]: indep_dummy_data['Intercept'] = 1.0 # add a column of 1's to the independent variables' dataframe so that the logit 

In [54]: # model will include intercept in the fitting process.  

In [55]: indep_dummy_data.head(5) # take a look at the first five rows in the dataframe 
Out[55]: 
   Tenure  TotalCharges  ...  PaymentMethod_Mailed check  Intercept
0       1         29.85  ...                           0        1.0
1      34       1889.50  ...                           1        1.0
2       2        108.15  ...                           1        1.0
3      45       1840.75  ...                           0        1.0
4       2        151.65  ...                           0        1.0

[5 rows x 9 columns]

In [56]: # Note: to use sm to estimate a logit model, the dependent variable (Creditability) MUST be 0s and 1s. Other levels cannot be used.

In [57]: sm_logit_model = sm.Logit(Telecom_Churn_Data['Churn'], indep_dummy_data)

In [58]: # note that the format of the fitting specification is sm.Logit(y,X) where X is the independent variables' matrix, 

In [59]: # and y is the dependent variable.

In [60]: sm_fitting_results = sm_logit_model.fit() # fit the logit model
Optimization terminated successfully.
         Current function value: 0.434784
         Iterations 8

In [61]: sm_fitting_results.summary() # you can also output the whole results, including the Pseudo R-square and the log-likelihood value
Out[61]: 
<class 'statsmodels.iolib.summary.Summary'>
"""
                           Logit Regression Results                           
==============================================================================
Dep. Variable:                  Churn   No. Observations:                 7031
Model:                          Logit   Df Residuals:                     7022
Method:                           MLE   Df Model:                            8
Date:                Wed, 22 Jun 2022   Pseudo R-squ.:                  0.2492
Time:                        17:33:51   Log-Likelihood:                -3057.0
converged:                       True   LL-Null:                       -4071.4
Covariance Type:            nonrobust   LLR p-value:                     0.000
=========================================================================================================
                                            coef    std err          z      P>|z|      [0.025      0.975]
---------------------------------------------------------------------------------------------------------
Tenure                                   -0.0839      0.005    -15.485      0.000      -0.095      -0.073
TotalCharges                              0.0007    5.3e-05     12.568      0.000       0.001       0.001
Contract_One year                        -0.9797      0.103     -9.541      0.000      -1.181      -0.778
Contract_Two year                        -1.9865      0.172    -11.572      0.000      -2.323      -1.650
PaperlessBilling_Yes                      0.5875      0.071      8.300      0.000       0.449       0.726
PaymentMethod_Credit card (automatic)    -0.1070      0.111     -0.966      0.334      -0.324       0.110
PaymentMethod_Electronic check            0.5060      0.091      5.563      0.000       0.328       0.684
PaymentMethod_Mailed check               -0.3871      0.109     -3.565      0.000      -0.600      -0.174
Intercept                                -0.2896      0.102     -2.835      0.005      -0.490      -0.089
=========================================================================================================
"""

In [62]: # We need to call the following module:

In [63]: from sklearn.model_selection import train_test_split

In [64]: # Similar to R, let's set a random seed sp that each of us will select the same training and test sub-data 

In [65]: # for reproducing the same forecasting result.

In [66]: np.random.seed(1)

In [67]: from sklearn.model_selection import train_test_split

In [68]: # Similar to R, let's set a random seed sp that each of us will select the same training and test sub-data 

In [69]: # for reproducing the same forecasting result.

In [70]: np.random.seed(1)

In [71]: train_X, test_X, train_y, test_y = train_test_split(indep_dummy_data,Telecom_Churn_Data['Churn'], 
  File "C:\Users\rao8r\AppData\Local\Temp\ipykernel_9760\513116906.py", line 1
    train_X, test_X, train_y, test_y = train_test_split(indep_dummy_data,Telecom_Churn_Data['Churn'],
                                                                                                      ^
SyntaxError: unexpected EOF while parsing


In [72]: np.random.seed(1)
    ...: train_X, test_X, train_y, test_y = train_test_split(indep_dummy_data,Telecom_Churn_Data['Churn'], 
    ...: train_size=0.8, random_state = 0)

In [73]: train_X, test_X, train_y, test_y = train_test_split(indep_dummy_data,Telecom_Churn_Data['Churn'],train_size=0.8, random_state = 0)

In [74]: # above code line creates the training data and test data in one process

In [75]: # here train_X is the independent variables' training data; 

In [76]: # test_X is the independent variables' test data;

In [77]: # train_y is the dependent variable's training data;

In [78]: # test_y is the dependent variable's test data

In [79]: sm_logit_model_for_pred = sm.Logit(train_y, train_X)  # specify the logit model

In [80]: fitting_results_for_pred = sm_logit_model_for_pred.fit() # fit the logit model using the training data
Optimization terminated successfully.
         Current function value: 0.432035
         Iterations 8

In [81]: fitting_results_for_pred.summary() # you can also output the estimation results, including the Pseudo R-square and the log-likelihood value
Out[81]: 
<class 'statsmodels.iolib.summary.Summary'>
"""
                           Logit Regression Results                           
==============================================================================
Dep. Variable:                  Churn   No. Observations:                 5624
Model:                          Logit   Df Residuals:                     5615
Method:                           MLE   Df Model:                            8
Date:                Wed, 22 Jun 2022   Pseudo R-squ.:                  0.2553
Time:                        17:37:10   Log-Likelihood:                -2429.8
converged:                       True   LL-Null:                       -3262.7
Covariance Type:            nonrobust   LLR p-value:                     0.000
=========================================================================================================
                                            coef    std err          z      P>|z|      [0.025      0.975]
---------------------------------------------------------------------------------------------------------
Tenure                                   -0.0836      0.006    -13.821      0.000      -0.095      -0.072
TotalCharges                              0.0007   5.92e-05     11.173      0.000       0.001       0.001
Contract_One year                        -0.9723      0.116     -8.415      0.000      -1.199      -0.746
Contract_Two year                        -2.0233      0.194    -10.406      0.000      -2.404      -1.642
PaperlessBilling_Yes                      0.5904      0.080      7.406      0.000       0.434       0.747
PaymentMethod_Credit card (automatic)    -0.1704      0.125     -1.360      0.174      -0.416       0.075
PaymentMethod_Electronic check            0.5314      0.102      5.187      0.000       0.331       0.732
PaymentMethod_Mailed check               -0.4225      0.123     -3.446      0.001      -0.663      -0.182
Intercept                                -0.2795      0.116     -2.416      0.016      -0.506      -0.053
=========================================================================================================
"""

In [82]: pred_y = round(fitting_results_for_pred.predict(test_X)) # make forecasts for the dependent variable using the trained logit model

In [83]: # here round() converts the predicted probability to either 1 or 0.

In [84]: # Construct a confusion matrix

In [85]: confusion_matrix = pd.crosstab(test_y, pred_y, rownames=['Actual y'], colnames=['Predicted y'])  

In [86]: # or we can directly output the accuracy rate

In [87]: from sklearn.metrics import accuracy_score

In [88]: accuracy_rate = accuracy_score(test_y, pred_y)

In [89]: print("The accuracy rate of the forecasting is: ",accuracy_rate)
The accuracy rate of the forecasting is:  0.7825159914712153

In [90]: