Churn_Modelling in Python

By Kotha Sai Narasimha Rao

The main aim of this Python jupyter project is to create a job demographic segmentation model to tell the bank which of its customers are at the highest risk of leaving.

Dataset Link: https://www.kaggle.com/aakash50897/churn-modellingcsv

About the dataset :

The dataset contains 10,000 rows and 14 columns. Out of 14 features, 13 features are independent features and 1 is a dependent feature. The main task is to find the customers who are at the highest risk of leaving or the customer is reliable to the bank based on the customer data present in the bank and that could govern the bank's decision whether or not to give loans.

Dataset description :

1)CustomerId: Unique id of the customer in the bank.

2)Surname Name of the customer.

3)CreditScore: The credit score of the customer.

4)Geography: It tells the area.

5)Gender: It tells about whether the customer is male or female.

6)Age: It describes the age of the customer.

7)Tenure: How long the customer have been in the bank.

8)Balance: The bank balance of the respective customer at that time.

9)No of products: It tells about the no of products are there for the customer.

10)HasCrCard: it tells whether the customer has a credit card or not.

11)IsActiveMember: it tells whether the customer is an active member or not.

12)EstimatedSalary: it tells the estimated salary of the customer based on the prior data.

13)Exited: It tells whether the customer is excited from the bank or not.

1-excited from the bank.

0-Not excited from the bank.

Data Preprocessing and Data Visualization

Step 1:

Importing the required Python Libraries

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.preprocessing import LabelEncoder
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import OneHotEncoder
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
import tensorflow as tf
from sklearn.metrics import classification_report

Step 2:

Importing  the dataset

data= pd.read_csv(r'P:\Churn_Modelling.csv')
data.head()

Data.head() commands prints the first five rows of the dataset

Step 3:

data.info()

Output :

#   Column           Non-Null Count  Dtype  
---  ------           --------------  -----  
 0   RowNumber        10000 non-null  int64  
 1   CustomerId       10000 non-null  int64  
 2   Surname          10000 non-null  object 
 3   CreditScore      10000 non-null  int64  
 4   Geography        10000 non-null  object 
 5   Gender           10000 non-null  object 
 6   Age              10000 non-null  int64  
 7   Tenure           10000 non-null  int64  
 8   Balance          10000 non-null  float64
 9   NumOfProducts    10000 non-null  int64  
 10  HasCrCard        10000 non-null  int64  
 11  IsActiveMember   10000 non-null  int64  
 12  EstimatedSalary  10000 non-null  float64
 13  Exited           10000 non-null  int64  

Step 4:

Checking the Null values in the dataset

data.isnull().sum()

Output :

RowNumber          0
CustomerId         0
Surname            0
CreditScore        0
Geography          0
Gender             0
Age                0
Tenure             0
Balance            0
NumOfProducts      0
HasCrCard          0
IsActiveMember     0
EstimatedSalary    0
Exited             0
dtype: int64

Step 5:
Statistical analysis of the features in dataset.

data.describe()

The above chunk of code prints the mean, count, Maximum value, Minimum Value of each feature Present in the dataset.

Step 6:

Analyzing the Gender Variable (Getting no of classes in gender variable)

data['Gender'].value_counts()

Output:

Male      5457
Female    4543
Name: Gender, dtype: int64

Step 7:
Comparision of Male and Female with respect to their Frequency using bar plot.

classes = pd.value_counts(data['Gender'], sort = True)

classes.plot(kind = 'bar', rot=0)

plt.title('comparison of male and female')
plt.xlabel('Gender')
plt.ylabel('population')
plt.show()

Step 8:

Analyzing the Age variable.

data['Age'].value_counts()

Output:

37    478
38    477
35    474
36    456
34    447
     ... 
92      2
88      1
82      1
85      1
83      1
Name: Age, Length: 70, dtype: int64

Step 9:
Comparision of Age 

plt.hist(x = data.Age, bins = 10, color = 'orange')
plt.title('comparison of Age')
plt.xlabel('Age')
plt.ylabel('population')
plt.show()

Step 10 :

Finding the correlation between the variables using heatmap.

sns.heatmap(data.corr(), annot = True, vmin=-1, vmax=1, center= 0)

Step 11:

Splitting the Dataset column-wise into two parts.

X = data.iloc[:, 3:-1].values
y = data.iloc[:, -1].values

X-Independent Features

y-Dependent Feature.

Step 12:

Encoding the Categorical data

Label Encoding the "Gender" column

le = LabelEncoder()
X[:, 2] = le.fit_transform(X[:, 2])
print(X)

Output :

[[619 'France' 0 ... 1 1 101348.88]
 [608 'Spain' 0 ... 0 1 112542.58]
 [502 'France' 0 ... 1 0 113931.57]
 ...
 [709 'France' 0 ... 0 1 42085.58]
 [772 'Germany' 1 ... 1 0 92888.52]
 [792 'France' 0 ... 1 0 38190.78]]

Step 13:

One Hot Encoding the "Geography" column using the column Transformer.

ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [1])], remainder='passthrough')
X = np.array(ct.fit_transform(X))
print(X)

Output:

[[1.0 0.0 0.0 ... 1 1 101348.88]
 [0.0 0.0 1.0 ... 0 1 112542.58]
 [1.0 0.0 0.0 ... 1 0 113931.57]
 ...
 [1.0 0.0 0.0 ... 0 1 42085.58]
 [0.0 1.0 0.0 ... 1 0 92888.52]
 [1.0 0.0 0.0 ... 1 0 38190.78]]

Step 14:

Feature Scaling 

sc = StandardScaler()
X = sc.fit_transform(X)
print(X)

Output :

[[ 0.99720391 -0.57873591 -0.57380915 ...  0.64609167  0.97024255
   0.02188649]
 [-1.00280393 -0.57873591  1.74273971 ... -1.54776799  0.97024255
   0.21653375]
 [ 0.99720391 -0.57873591 -0.57380915 ...  0.64609167 -1.03067011
   0.2406869 ]
 ...
 [ 0.99720391 -0.57873591 -0.57380915 ... -1.54776799  0.97024255
  -1.00864308]
 [-1.00280393  1.72790383 -0.57380915 ...  0.64609167 -1.03067011
  -0.12523071]
 [ 0.99720391 -0.57873591 -0.57380915 ...  0.64609167 -1.03067011
  -1.07636976]]

Step 15:
Splitting the dataset into Training set and testing set.

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)

Model Building :

Building the Artificial Neural Network(ANN):

Steps for Building the ANN:

1)Initializing the ANN.

2)Adding the Input layer and the First Hidden layer.

3)Adding the Second Hidden layer.

4)Adding the Output layer.

ann = tf.keras.models.Sequential()
ann.add(tf.keras.layers.Dense(units=6, activation='relu'))
ann.add(tf.keras.layers.Dense(units=6, activation='relu'))
ann.add(tf.keras.layers.Dense(units=1, activation='sigmoid'))

Compiling the ANN model

ann.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])

Training the ANN model on the training set

ann.fit(X_train, y_train, batch_size = 32, epochs = 100)

Predicting the output

y_pred = ann.predict(X_test)
y_pred = (y_pred > 0.5)

Evaluating the Model

_{True Positive: when a case was positive and predicted positive.} 

_{True Negative: When a case was negative and predicted negative.}

_{False Positive: when a case was negative and predicted was positive.} 

_{False Negative: when a case was positive and predicted was Negative.}

Confusion Matrix :

from sklearn.metrics import confusion_matrix, accuracy_score
confusion_matrix= confusion_matrix(y_test, y_pred)
print(confusion_matrix)

Output :

[[1519   76]
 [ 203  202]]

accuracy_score(y_test, y_pred)

_{Classification Report For the ANN}

print(classification_report(y_test,y_pred))

Output :

precision    recall  f1-score   support

           0       0.88      0.95      0.92      1595
           1       0.73      0.50      0.59       405

    accuracy                           0.86      2000
   macro avg       0.80      0.73      0.75      2000
weighted avg       0.85      0.86      0.85      2000