Churn Prediction Project

Goal: predict customer churn risk with calibrated probabilities to target top-k customers for retention.

Dataset: IBM Telco Customer Churn (Telco-Customer-Churn.csv) — place in data/raw/ (download from Kaggle).

Next steps:

• Set up a Python environment and install dependencies: pip install -r requirements.txt.
• Run EDA in notebooks/01_eda.ipynb to inspect data, churn rate, missingness, and feature types.
• Build a baseline Logistic Regression pipeline in src/models/train.py.

Running EDA:

• From project root: cd /Users/tobiomotayo/Developer/ml/churn
• Activate venv: source .venv/bin/activate
• Launch Jupyter: jupyter notebook and open notebooks/01_eda.ipynb
• Outputs: summary saved to reports/eda_summary.md; figures to reports/figures/churn_rate.png and reports/figures/missingness_top10.png.

Repo structure (planned):

• data/raw/, data/processed/
• notebooks/
• src/data/, src/models/, src/serve/
• models/
• reports/figures/
• requirements.txt, todos.md