BANK CHURN ANALYSIS

franklin obiefule
Oct 28, 2024
4 min read

Updated: Nov 9, 2024

Link to Github:

https://github.com/franklintheanalyst1/franklintheanalyst1/blob/main/Solution%2BBank%2BChurn%2BData%20(1).xlsx

TOOL USED

EXCEL

DATASET SOURCE ATTRIBUTION

Shahriar's Sight Academy [UDEMY: Data Analysis Career Path; 72 Days of Data Analyst Bootcamp]

INTRODUCTION

This project aims to analyze churned data of 500 bank customers, gain insights and identify key factors contributing to customer churn and retention.

BACKGROUND/MOTIVATION

The dataset contains 500 bank customers with different customer behaviors. I was motivated to gain insights into these customer behaviors and identify factors that could lead to customer churn.

DATA COLLECTION

Dataset contains 500 bank customers. Key metrics include: Geography, Credit Score, Gender, Account Balance, Tenure, Active member, Number of Products used and Customer churn rate.

DATASET LIMITATIONS

Cleaning the raw data by replacing missing values and inconsistent data in 'Estimated Salary' and 'Balance' variables respectively with the average value.

ANALYSIS

Sample of raw data

I cleaned the raw data by replacing missing values and inconsistent data in 'Estimated Salary' and 'Balance' variables respectively with the average value.

Some analytical questions were solved using PIVOT TABLE

Calculate average score of each geography?
How the average account balance vary between genders within each country?
The distribution of Active versus Non-active members according to having a credit card?
Customer churn rate per number of products used?

Average credit score for customers who have exited compared to those who have stayed, across different tenure?

Research questions were solved using statistical tools DATA ANALYSIS TOOL PACK IN Excel

1. Is there any significant difference in customers' average earning between churned and not-churned?

T-test was used for this because the question describes comparing mean values of two groups(churned and not churned).

Since p-value(0.42) is greater than significance Level (α)[0.05], there is no evidence to reject the null hypothesis(H0). Therefore, there is no significant difference in customers' average earning between churned and not-churned.

2. Is there any significant difference in customers' average credit scores among customers' geographic location?

ANOVA[analysis of variance] was used for this question because it compares the means of three or more independent groups.(FRANCE, SPAIN & GERMANY).

Since p-value(0.37) is greater than significance Level (α)[0.05], there is no evidence to reject the null hypothesis(H0). Therefore, there is no significant difference in customers' average credit scores among customers' geographic location.

RESULTS

A. The average credit score for France, Germany and Spain are 592.55, 579.69 and 600.99 respectively with a gross average of 591.42.

B. The average account balance variation between genders within each country.

France(Female= $129,576 , Male= $133,204)

Germany(Female= $140,888, Male= $128,591)

Spain(Female=$120,986, Male=$119,667)

Gross Average = $128,389

C. Distribution of active members versus non-active members according to having a credit card.

Active members who do not have credit card =113

Active members who have credit card = 124

Non-active members who do not have credit card = 140

Non-active members who have credit card= 123

D. Customer churn rate per number of products used

272 churned customers have used at least 1 product.

228 non-churned customers have used at least 1 product.

E. Average credit score for customers who have exited compared to those who have stayed across different tenure.

For customers who have exited across different tenure, their average credit score is 587.73.

For customers who have stayed across different tenure, their average credit score is 595.81.

F. Is there any significant difference in customers' average earning between churned and not-churned?

T-test was used for this because the question describes comparing mean values of two groups(churned and not churned).

G. Is there any significant difference in customers' average credit scores among customers' geographic location?

ANOVA[analysis of variance] was used for this question because it compares the means of three or more independent groups.(FRANCE, SPAIN & GERMANY).

CONCLUSION

Factors that could lead to Churn:

1. Customer demographics(Age & Gender)- Younger customers may be more likely to switch banks for better offers. Different preferences and behaviors based on gender (eg geography of gender) may influence churn.

2. Account characteristics(Tenure and Account Balance)- Customers with shorter tenure may be more likely to churn. Low account balances could indicate dissatisfaction or lack of engagement.

3. Banking Products(Number of banking products/Has credit card)- Customers with fewer products may feel less attached to the bank. Lack of credit card might indicate limited engagement with the bank.

4. Customer engagement(Active member & Customer service experience)- Customers who are not actively using their accounts may be more likely to leave. Poor experiences can lead to dissatisfaction leading to churn.

5. Financial factors(Credit Score & Estimated salary)- A low credit score may lead to higher fees or restrictions prompting churn. High income earning customers may seek better banking options elsewhere.

Identifying Factors for Customer Retention

1. Surveys and Feedbacks- Conduct surveys to gather direct feedback from customers about their experiences and reasons for leaving.

2. Retention strategies(Targeted marketing & Improving Customer experience)- Create personalized offers for customers at risk of churning based on identified factors. Enhance services and support for customers showing signs of dissatisfaction.

REFERENCES

Shahriar's Sight Academy [UDEMY: Data Analysis Career Path; 72 Days of Data Analyst Bootcamp] edX: Customer Analytics.
"Data Science for Business" by Foster Provost and Tom Fawcett