![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F26097785%2F116b05cf323f8efa040fa3f9b6830adc%2FStats%20(3).jpg?generation=1772057862263667&alt=media)

## 🎯 Competition Objective

You are provided with historical customer data to develop a **predictive churn model**.

---

## 📌 Your Tasks

### 1️⃣ Model Development

* Train a **binary classification model** using `train.csv`.

### 2️⃣ Probability Prediction

* Generate churn probability predictions for `test.csv`.

### 3️⃣ Submission Format

* Submit **continuous probability scores** for each customer.

> ⚠️ The output must represent probabilities (e.g., 0.73, 0.18, 0.91).

---

## ❗ Important Guidelines

### 🔹 Probability Interpretation

* A **higher probability** indicates a **higher churn risk**.
* Example:

  * `0.85` → Very high churn likelihood
  * `0.12` → Low churn likelihood

### 🔹 Binary Outputs Are Not Allowed

* ❌ Submitting only `0` or `1` predictions is **not acceptable**.
* ✅ Continuous probability values are required.

### 🔹 Evaluation Criteria

* The competition evaluation focuses on **ranking performance**, not classification threshold accuracy.
* Models should prioritize **correct ordering of churn risk** rather than fixed cutoff classification.

---