Bank Customer Churn Prediction with Explainable ML Using Random Forest, AdaBoost, SVM & Interpretability Tools like LIME, PDP, and Permutation Importance
This project focuses on predicting customer churn for a bank using advanced machine learning models and enhancing model interpretability with Explainable ML techniques. The goal is not only to achieve high prediction accuracy but also to understand the "why" behind the predictions, which is crucial for decision-making in high-stakes environments like finance.
Key Features Churn prediction using Random Forest, AdaBoost, and Support Vector Machines
Data preprocessing, balancing using SMOTE, and pipeline-based modelling
Integration of LIME, Partial Dependence Plots (PDP), and Permutation Feature Importance for model explainability
Evaluation using F1-score, Accuracy, and Precision-Recall AUC
Exploratory Data Analysis (EDA) with Seaborn, Matplotlib, and Plotly
Results -
| Model | F1 Score | Accuracy |
|---|---|---|
| Random Forest | 0.98 | 0.98 |
| AdaBoost | 0.93 | 0.92 |
| SVM | 0.95 | 0.95 |
Explainable AI Tools Used LIME Provides local instance-level explanations. Example: being single or having a graduate education can affect churn probability.
Permutation Feature Importance Reveals most influential features:
Total_Trans_Ct
Total_Trans_Amt
Total_Relationship_Count
PDP (Partial Dependence Plot) Shows global feature impact; e.g., Dependent_count and Total_Revolving_Bal significantly affect churn.