Devdit/Datathon_CatBoost

🐱 CatBoost Models for Churn, Tenure, and LTV Prediction

This repository contains three CatBoost models trained to predict:

• Churn (clf_churn.pkl) – Binary classification (likelihood of customer churn)
• Tenure (RegTenure.pkl) – Regression (expected number of months a customer stays)
• Lifetime Value (LTV) (reg_ltv.pkl) – Regression (predicted total value of a customer)

Each model is saved using Python's pickle module and can be loaded easily for inference.

---

🧠 Model Overview

Model File	Task	Type
clf_churn.pkl	Churn Prediction	Classification
RegTenure.pkl	Tenure Estimation	Regression
reg_ltv.pkl	LTV Prediction	Regression

---

💾 How to Use

1. Install Requirements

pip install catboost pandas


import pickle

with open("clf_churn.pkl", "rb") as f:
    clf_cb = pickle.load(f)

with open("RegTenure.pkl", "rb") as f:
    reg_tenure_cb = pickle.load(f)

with open("reg_ltv.pkl", "rb") as f:
    reg_ltv_cb = pickle.load(f)


# Predict churn probability
churn_proba = clf_cb.predict_proba(X_test)[:, 1]

# Predict tenure
tenure_pred = reg_tenure_cb.predict(X_test)

# Predict lifetime value
ltv_pred = reg_ltv_cb.predict(X_test)

print("🔁 Churn:", churn_proba[:5])
print("📅 Tenure:", tenure_pred[:5])
print("💰 LTV:", ltv_pred[:5])