Churn Prediction with Explainability (SHAP)

Predict churn with logistic regression + gradient boosting, then explain every prediction with SHAP so the model is decision-ready.

Domain: business_analytics · Relevant roles: Data Analyst, Product Analyst, Business Intelligence Analyst

Problem

Why this matters and the decision it informs.

Data

Source: Telco Customer Churn (Kaggle/IBM) Place raw files in data/ (git-ignored). Document how to obtain them here.

Approach

1. Data cleaning & EDA
2. Modeling / analysis
3. Validation & interpretation
4. Business recommendation

Key results

Add the headline chart and the one-sentence takeaway here (this is what recruiters read).

Tech stack

Python, XGBoost, SHAP, classification

How to run

pip install -r requirements.txt
jupyter notebook notebooks/01_analysis.ipynb

---

Part of Ela Kumuk's analytics portfolio.