sushantsriv / customer-churn-prediction

Objective: To develop a model that can predict with high accuracy whether a customer will continue using the bank's services or leave (churn).

Churn Prediction Analysis for Retail Banking

Introduction

This repository contains the machine learning models and analysis for predicting customer churn in retail banking. The objective is to leverage data to predict whether a customer will stay with the bank or leave (churn) with high accuracy.

Dataset

The dataset used for this analysis is a dummy dataset with 10,000 rows containing various demographic and banking information for customers, with no missing values.

Analysis Overview

• Exploratory Data Analysis (EDA): To understand the distribution and relationships within the data.
• Feature Engineering: Creation of additional features like Credit Score to Age Ratio and Balance to Salary Ratio for better insights.
• Class Imbalance Handling: Use of SMOTE (Synthetic Minority Over Sampling Technique) to balance the dataset.
• Model Building: Development of multiple predictive models including SVM (with and without SMOTE), Random Forest, and Neural Networks.
• Model Evaluation: Models are evaluated based on their accuracy and the rate of false negatives.
• Inferences: Derived from plots and model outcomes to understand the characteristics of churning and remaining customers.

Results

The Neural Network model achieved the best performance with an accuracy of around 86% and a false negative rate of 2.11%, indicating a high predictive capability for actual churn events.

Recommendations

Product Portfolio Evaluation

• Assess and refine product offerings.

Credit Score and Tenure Analysis

• Further explore the lack of distinctions in credit score and tenure between churned and retained customers through customer feedback surveys.

Balance Retention Strategies

• Develop personalized strategies to retain customers with significant account balances.

Regular Monitoring and Adaptation

• Implement continuous monitoring of strategy effectiveness.

Collaboration with Customer Support

• Enhance the collaboration between data analysis teams and customer support.

Promotion of Financial Literacy

• Launch educational campaigns for customers to make informed financial decisions.

Incentivize Customer Feedback

• Encourage and utilize customer feedback for improvements.

Cross-Functional Team Collaboration

• Promote collaboration between various teams to enhance customer satisfaction and retention.