## 📜 1 – Purpose & Honor Code

This is a skills-based data-science exercise for **COURSE NAME**.
By entering you agree to:

* **Play fair.** One Kaggle account per person or team.
* **Work independently.** Share insights, not private test labels or predictions.
* **Show your work.** After the deadline you must upload to the course classroom (Google Classroom) a notebook or script that reproduces your best public-leaderboard score **without internet access**. Scores that can’t be reproduced lose credit.
* **No “shortcut” data.** External data are allowed only if they are produced by you.
* **You can't use any neural network model exept for data augmentation or you won't get any score
* **you must upload a completely runnable code in classroom which produces your submition file.
Violation of the honor code = disqualification and zero points.

---

## 📤 2 – Submissions

| Limit                                                | Value                                           |
| ---------------------------------------------------- | ----------------------------------------------- |
| **Daily submissions**                                | up to **20** per team                           |
| **Total “private” submissions re-scored at the end** | your **best 20** (the last 20 if you made more) |

*Kaggle shows your public score on 50% of the hidden labels. Final ranking uses the remaining 50%.*

---

## 🏅 3 – Grading & Prizes

* **Metric:** overall accuracy on the private leaderboard.
* **Course credit:** your grade is based on private-leaderboard rank; your bonus score will be calculated as
$$
\text{Bonus} \;=\; \frac{1}{1 + 0.5 * (\text{rank} - 1)}
$$

All winning teams must share a short write-up and cleaned code with the class.

---


### Need Help?

you can message TAs anytime.
**Have fun and good luck!**

---