# Challenge 02 – Errors -> Integrity ( 2026 ) — Data

来源：https://www.kaggle.com/competitions/challenge-02-errors-integrity-2026/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

## 🧱 1. Overall structure target

You want the data tab for Challenge 02 to have **three layers**, like Challenge 01:

1. **Dataset Description block** – a clean narrative summary (1–2 screens max)
2. **Files table** – already auto-generated by Kaggle
3. **About this directory** – a short sidebar paragraph (optional but recommended)
4. **README.txt** – downloadable, mirrors the Dataset Description text for offline viewing

Right now, Challenge 02 has a *full technical README inside the Dataset Description*, which makes the page long and redundant.

---

## 🧩 2. What to change in Kaggle

| Section                                       | What to do                                                                                                  | Why                                             |
| --------------------------------------------- | ----------------------------------------------------------------------------------------------------------- | ----------------------------------------------- |
| **Dataset Description (main editable block)** | Replace the long markdown README with a concise version like below. Keep headings 🧹 and 📁 for continuity. | Matches Challenge 01’s tone and length.         |
| **README.txt file**                           | Keep the detailed technical README you already wrote — that becomes the downloadable file.                  | Keeps depth without cluttering the web layout.  |
| **About this directory (sidebar)**            | Add a short 3-line summary (I’ll provide it).                                                               | Gives context in file panel, like Challenge 01. |

---

## ✍️ 3. Replacement text for the main “Dataset Description” box

You can paste this directly in Kaggle’s **Edit Dataset Description**:

---

### 🧹 Dataset Description

This dataset accompanies **Challenge 02 – Errors → Integrity** in the *Grocery Retail – Driven by Data* series.
It provides a structured but imperfect grocery database — complete yet noisy — for practicing systematic **data auditing, repair, and validation**.

#### 📁 Files Included

| File             | Description                                                               |
| ---------------- | ------------------------------------------------------------------------- |
| `products.csv`   | Product master list with IDs, names, and vendor links.                    |
| `vendors.csv`    | Supplier reference table containing duplicates and mismatched keys.       |
| `stores.csv`     | Store metadata with regional and time-zone variations.                    |
| `sales.csv`      | Transaction-level data with missing totals and duplicates for validation. |
| `categories.csv` | Product category hierarchy, with a few orphaned entries.                  |
| `README.txt`     | Full technical notes and workflow guide.                                  |

#### 🧾 Objective

Audit, repair, and validate the data to build a **trustworthy foundation** for analytics and forecasting in later challenges.
Participants identify missing or inconsistent values, apply principled fixes, and document every change transparently.

#### ⚖️ License

**CC BY-NC-SA 4.0** — free for non-commercial educational and research use with attribution.