Pens & Pencils – AutoML Best Model (LightGBM)

Model Details

• Algorithm: LightGBM (bagged ensemble, level 1)
• Framework: AutoGluon Tabular
• Task: Binary classification
• Target: is_pen (1 = pen, 0 = pencil)

Data

• Origin: Collected manually by classmate, augmented with synthetic rows (label-preserving jitter).
• Features (6): length_cm, weight_g, is_metallic, body_shape, has_clip, is_pen.
• Splits: Original (30 rows), augmented synthetic (300 rows).

Training Setup

• Best model: LightGBM_r161_BAG_L1
• Hyperparameters: See training notebook (depth, learning rate, etc.).
• Hardware: Google Colab (CPU/GPU as available).

Metrics

• Accuracy: 1.00

Limitations & Ethical Notes

• Dataset is small and synthetic augmentation may bias results.
• Not suitable for real-world manufacturing or product design.
• Educational use only.

License

AI Usage Disclosure

• Notebook partially assisted by GenAI (ChatGPT) for code refactoring and documentation.
• Core modeling and evaluation by AutoGluon.