![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F26097785%2F891a736132476ceb4fb11953b878ae2a%2FStats%20(8).jpg?generation=1772058486843117&alt=media)


# 🚫 Prohibited Practices & Compliance Rules

To ensure fairness, reproducibility, and methodological integrity, the following practices are **strictly prohibited**.

---

## ❌ Disallowed Activities

### 🔹 Data Leakage (Any Form)

* Using future information during training
* Encoding target information improperly
* Fitting preprocessing steps on combined train + test data
* Any cross-fold contamination

> Any evidence of leakage will result in immediate disqualification.

---

### 🔹 Manual Modification of Prediction Files

* Editing `submission.csv` manually
* Adjusting probabilities post-training
* Threshold tuning directly inside the submission file

All predictions must be generated programmatically.

---

### 🔹 Hardcoded Outputs

* Predefined probability values
* Static or fabricated prediction arrays
* Conditional logic that bypasses model inference

The notebook must compute predictions through a trained model.

---

### 🔹 Use of External Datasets

* No external data sources
* No pretrained models trained on external customer data
* No enrichment from third-party APIs

All modeling must rely exclusively on the provided competition data.

---

### 🔹 Reverse-Engineering the Test Set

* Attempting to infer hidden labels
* Exploiting leakage patterns in public leaderboard feedback
* Any manipulation aimed at uncovering private test targets

---

## 📊 Submission Limit

Participants are limited to:

> **5 submissions per day**

This encourages thoughtful experimentation rather than leaderboard probing.

---

## ⚖️ Enforcement

Violation of any of the above rules may result in:

* Score invalidation
* Submission removal
* Full disqualification from the competition

---