Business Report on The Thera bank credit cards

Table of Content
Business Context - Problem Statement 3
Key Features for Analysis 4-6 Pages
1
 Types of charts 7-23 Pages
Business Insights: 24

Business Context – Problem Statement

The Thera bank recently saw a steep decline in the number of users of their credit card,
credit cards are a good source of income for banks because of different kinds of fees
charged by the banks like annual fees, balance transfer fees, and cash advance fees, late
payment fees, foreign transaction fees, and others. Some fees are charged to every user
irrespective of usage, while others are charged under specified circumstances.

2
 Customers’ leaving credit cards services would lead bank to loss, so the bank wants to
analyse the data of customers and identify the customers who will leave their credit card
services and reason for same – so that bank could improve upon those areas

You as a Data scientist at Thera bank need to come up with a classification model that will
help the bank improve its services so that customers do not renounce their credit cards

Objective: As a data scientist at Thera Bank, my task is to create a classification model to

predict whether a customer will leave the bank's credit card service (i.e., become an
"Attrited Customer"). The aim is to identify the factors that drive customers to stop using
their credit cards, which will help the bank take corrective actions to retain customers.

Key Features for Analysis:

1.Demographic Information:
 Customer_Age: The age of the account holder.
 Gender: The gender of the account holder.
 Dependent_count: Number of dependents.
 Education_Level: Educational qualification (Graduate, College, Doctorate, etc.).
 Marital_Status: Marital status of the customer.
 Income_Category: Annual income category of the customer.

3
2.Banking Relationship:
 Months_on_book: Duration of the customer’s relationship with the bank.
 Total_Relationship_Count: Number of products (services) the customer has with the
bank.
3.Credit Card Activity:
 Months_Inactive_12_mon: Number of months the customer was inactive in the past
12 months.
 Contacts_Count_12_mon: Number of contacts made with the customer in the past 12
months.
4.Credit and Usage Information:
 Credit_Limit: Total credit limit on the customer’s credit card.
 Total_Revolving_Bal: The balance that the customer did not pay in full, carried over
from month to month.
 Avg_Open_To_Buy: Amount left on the credit card for spending (on average over the
last 12 months).
 Avg_Utilization_Ratio: The percentage of the credit limit used by the customer
(calculated using the relationship between credit limit and revolving balance).
 Total_Amt_Chng_Q4_Q1: Change in transaction amount from Q4 to Q1.
 Total_Trans_Amt: Total transaction amount in the last 12 months.
 Total_Trans_Ct: Total number of transactions in the last 12 months.
 Total_Ct_Chng_Q4_Q1: Change in transaction count from Q4 to Q1.

Steps for Analysis:

1.Data Preprocessing:
 Handle missing values in categorical features like Education_Level or
Income_Category.
 Convert categorical features (e.g., Gender, Education_Level, Marital_Status) to
numerical values for modeling.

4
2.Exploratory Data Analysis (EDA):
 Explore correlations between key features such as credit utilization, transaction
patterns, and customer attrition.
 Analyze the distribution of customer demographics (age, income, etc.) and credit card
usage.
3.Model Selection:
 Use classification algorithms (e.g., Logistic Regression, Random Forest, Gradient
Boosting) to predict whether a customer will leave.
 Evaluate model performance using metrics such as accuracy, precision, recall, and F1-
score to ensure proper detection of attrited customers.

4.Feature Importance:
 Identify the most significant predictors of customer attrition, such as inactivity, high
revolving balances, or a decrease in transaction amounts.

5.Actionable Insights:

 Based on the model results, the bank can take proactive measures, such as offering
better credit terms, sending reminders to inactive customers, or creating personalized
offers to retain valuable customers.

Type of Data

The data types: float64(5), int64(10), object(6)

The dataset contains 10,127 records and 21 columns. Here's a summary of the key
features:

Data Dictionary

 CLIENTNUM: Unique identifier for the customer.

5
  Attrition_Flag: Indicates whether the customer is an "Existing Customer" or "Attrited
Customer".
 Customer_Age: Age of the customer.
 Gender: Gender of the customer.
 Dependent_count: Number of dependents.
 Education_Level: Educational background (some missing values).
 Marital_Status: Marital status (some missing values).
 Income_Category: Annual income category.
 Card_Category: Type of credit card.
 Months_on_book: Duration of the customer relationship with the bank.
 Total_Relationship_Count: Number of products/services held by the customer.
 Months_Inactive_12_mon: Number of months the customer was inactive in the last
12 months.
 Contacts_Count_12_mon: Number of contacts with the customer in the last 12
months.
 Credit_Limit: Credit limit on the card.
 Total_Revolving_Bal: The revolving balance.
 Avg_Open_To_Buy: Average amount left to spend on the card.
 Total_Amt_Chng_Q4_Q1: Change in transaction amount between Q4 and Q1.
 Total_Trans_Amt: Total transaction amount in the last 12 months.
 Total_Trans_Ct: Total number of transactions in the last 12 months.
 Total_Ct_Chng_Q4_Q1: Change in transaction count between Q4 and Q1.
 Avg_Utilization_Ratio: Average utilization of the available credit.

Data Considerations:
 Missing values in the Education_Leveland Marital_Statuscolumns need handling.
 Attrition_Flagis the target column, which will be used to predict whether a
customer will leave.
The dataset has missing values in the following columns:
 Education_Level: 1,519 missing values.
 Marital_Status: 749 missing values.

Data Analysis
 Mean value for theCustomer Agecolumn is approx 46 and the median is also 46. This
shows that majority of the customers are under 46 years of age.
 Dependent Countcolumn has mean and median of~2

6
 Months on Bookcolumn has mean and median of36months.Minimumvalue is 13
months, showing that the dataset captures data for customers with the bank at least
1 whole years
 Total Relationship Counthas mean and median of~4
 Credit Limithas a wide range of1.4K to 34.5K, the median being4.5K, way less than
the mean8.6K • Total Transaction Counthas mean of~65and median of67

7
Months_on_book

Credit_Limit

8
Total_Revolving_Bal

Avg_Open_To_Buy

9
Total_Trans_Ct

Total_Amt_Chng_Q4_Q1

10
Let's see total transaction amount distributed
Total_Trans_Amt

Total_Ct_Chng_Q4_Q1

11
Avg_Utilization_Ratio

Total_Relationship_Count

12
Months_Inactive_12_mon

Contacts_Count_12_mon

13
Gender

Let's see the distribution of the level of education of customers

Education_Level

14
Marital_Status

15
Let's see the distribution of the level of income of customers
Income_Category

16
Card_Category

17
Attrition_Flag

18
Bivariate Distributions
Let's see the attributes that have a strong correlation with each other
Correlation Check
_Flag 0 1 All
Gender
All 8500 1627 10127
F 4428 930 5358
M 4072 697 4769

19
Attrition_Flag vs Education_Level
Education_Level College Doctorate Graduate High School Post-Graduate \
Attrition_Flag
All 1013 451 3128 2013 516
0 859 356 2641 1707 424
1 154 95 487 306 92

Education_Level Uneducated All

Attrition_Flag
All 1487 8608
0 1250 7237
1 237 1371

20
Attrition_Flag vs Income_Category
Income_Category $120K + $40K - $60K $60K - $80K $80K - $120K \
Attrition_Flag
All 727 1790 1402 1535
0 601 1519 1213 1293
1 126 271 189 242

Income_Category Less than $40K abc All

Attrition_Flag
All 3561 1112 10127
0 2949 925 8500
1 612 187 1627
------------------------------------------------------------------------------
------------------------------------------

21
Attrition_Flag vs Contacts_Count_12_mon
------------------------------------------------------------------------------

-----------

Let's see the number of months a customer was inactive in the last 12 months
(Months_Inactive_12_mon) vary by the customer's account status (Attrition_Flag)

22
Attrition_Flag vs Credit_Limit

23
Attrition_Flag vs Customer_Age

Actionable Insights & Recommendations

1.Attrition Rate: Understanding the proportion of attired vs. existing customers is key to
measuring the scale of the issue.
2.Customer Segmentation: By segmenting customers based on their age, income, marital
status, and education, the bank can identify which groups are more likely to leave and tailor
services accordingly.
3.Credit Usage: Customers with high revolving balances, low credit utilization ratios, or
large variations in transaction activity might be at higher risk of attrition due to
dissatisfaction with the product offering or financial strain.
4.Customer Inactivity: Customers inactive for many months or with fewer contacts in a year
may indicate a weak engagement level, increasing the likelihood of them leaving.
 The most important features to understand customer credit card churn, are
 Total Transaction Count
 Total Transaction Amount
 Total Revolving Balance
 Total Amount Change Q4 to Q1
 Total Count Change Q4 to Q1
 Total Relationship Count.
 All of these features are negatively correlated with the Attrition Flag, meaning, the
lower the values of these features, the higher the chances of a customer to attrite.
 Bank should connect with the customer more often to increase the connect, and
provide the customer with various offers and schemes to increase relationships of
the customer with the bank.
 Bank should offer cashback schemes on credit cards, which might encourage the
customers on using the credit card more often
 Bank should also offer credit limit increase for the customers who are regularly using
the credit card. This should increase the credit card spends/transaction amounts.
 0% interest EMI on credit card is also another offer that can be provided to
customers to encourage the customers buy products of higher cost with credit card,
and convert the expenditure to EMI, so that total transaction amount as well as
transaction counts would increase. The balance would also revolve nicely.

24
 Along with the available types of cards, bank can introduce credit cards specific to
online shopping (with % cashback offers) or online food ordering. This way the card
will be used more frequently. With our model, we can predict which customers are
likely to attrite, and according to the predicted probability, at least top 20-30%
customers can be reached out to discuss credit card offers, credit limit increase etc,
to try retain those customers.

25