ML Demo Models

This directory contains simple machine learning demo models for various well-known datasets, useful for portfolio or demo purposes.

Datasets Covered

Tabular / Classification

• Wine (UCI Wine) — 3 classes, chemistry features.
• Breast Cancer Wisconsin (Diagnostic) — binary cancer detection.
• Pima Indians Diabetes — binary classification, healthcare.
• Digits (sklearn.datasets.load_digits) — small image classification (8x8).
• Heart Disease (UCI) — risk prediction (binary).

Image

• Fashion-MNIST — 28x28 clothing items, good for CNN rookies.
• Cats vs Dogs (small subset) — classic; small subset for demo.

Text / Sentiment

• IMDb (small) — movie review sentiment.
• Sentiment140 — tweets with sentiment.

Music / Mood

• DEAM — song-level valence/arousal annotations.
• Million Playlist Dataset (Spotify) — subset for recommendation demo.
• RAVDESS — emotional speech/audio.
• GTZAN / FMA — genre datasets for audio feature extraction.

How to Run

1. Install dependencies: pip install -r requirements.txt
2. Run individual scripts: python <script_name>.py
3. Each script will train a simple model and display evaluation metrics.

Notes

• For large datasets like Sentiment140 or Cats vs Dogs, you may need to download subsets manually.
• Music/mood datasets require audio files; the demo provides feature extraction code.
• Adjust hyperparameters and model architectures as needed for your use case.