Update README.md

README.md

@@ -31,4 +31,73 @@ configs:
     path: data/original-*
   - split: augmented
     path: data/augmented-*
+license: cc
+language:
+- en
 ---
+
+# Model Card - Beverage Energy Tracker Dataset
+
+##  Purpose
+This dataset was created to analyze and model the energy related characteristics of common beverages based on measurable features. It is intended for educational and experimental machine learning use cases, such as binary classification and feature importance analysis.
+
+## Composition
+The dataset contains:
+- 30 manually collected real world samples in the `original` split
+- 300 label preserving synthetic samples in the `augmented` split  
+Each sample includes:
+- 5 input features (category, Added sugar(g), Calories, Volume, Energy rating)
+- 1 binary target variable indicating the sugar level (is high sugar?)
+
+## Collection
+Data was manually gathered from the product description pages of:
+- Starbucks  
+- ALDI USA  
+- MarketDistrict  
+- Amazon  
+- Walmart  
+
+Feature values such as sugar content, calories, volume and category were noted for each drink. Energy rating collected from consumers (Friends). The binary target (`is_high_sugar?`) was assigned based on sugar content (0 if <20g).
+
+## Preprocessing & Augmentation
+- Basic preprocessing: cleaned column names, ensured numerical features were floats/integers.
+- For augmentation, **label preserving jitter techniques** were used:
+  - Gaussian noise was added to continuous features (`Added sugar (g)`, `Calories`).
+  - ±1 encoding step changes applied to the integer feature (`Energy rating`), clipped to valid bounds (1-5).
+- Categorical fields (`Beverage type`, `Volume`) were left unchanged to preserve label consistency.
+- This process expanded the dataset to 300 augmented samples.
+
+
+## Labels
+- **Features:**
+  - `Beverage type` (categorical, string)
+  - `Added sugar (g)` (float)
+  - `Calories` (float)
+  - `Volume` (categorical, 300 mL or 355 mL)
+  - `Energy rating` (int, 1–5)
+- **Target:**  
+  - `is high sugar?` – integer label, also used for prediction tasks
+
+## Splits
+- `original`: 30 real world examples
+- `augmented`: 300 synthetic samples via jitter based augmentation
+
+## Intended Use / Limitations
+- **Use cases:** regression/classification practice, feature importance exploration, interpretable ML projects
+- **Limitations:** small original sample size, subjective energy rating label, dataset not exhaustive of all beverage categories
+
+This dataset is **not** intended for nutrition advice or commercial recommendation engines.
+
+## Ethical Notes
+- No personally identifiable information is included.
+- Energy ratings are subjective and approximate.
+- Consumers should not use this data for health or fitness decisions.
+
+## License
+This dataset is released under the **Creative Commons Attribution 4.0 International (CC BY 4.0)** license.
+
+## AI Usage Disclosure
+ChatGPT (GPT-4) was used as a **teammate throughout the process** to assist with:
+- On the go Python coding
+- Augmentation approach
+- Polishing and formatting the final dataset and notebook