Coursework experiments in classification, regression, game-playing, and tokenization.
These notebooks document problem framing, data preparation, model comparison, and interpretation. They are learning projects—not production models or independently validated research results.
- Email spam classification: an initial workflow for separating spam from legitimate email.
- San Francisco housing price prediction: data preparation, leakage-aware feature selection, and random-forest regression.
- Customer purchase and return classification: classification across purchase, return, and non-purchase outcomes.
- Email feature classification: text cleaning and engineered features for spam classification.
- Tetris game playing: experiments with a provided game framework.
- Wikipedia BPE tokenizer: text extraction, tokenizer training, and train/test preparation scripts.
Open notebooks on GitHub for saved explanations and outputs, or use Jupyter in a dedicated Python environment. Inspect imports, dataset paths, and setup cells before running: projects were developed in different course environments and do not share a pinned dependency set.
Some datasets and framework components were supplied for coursework. Their availability and reuse rights should not be assumed. Saved outputs describe a particular run, not a guarantee of reproducible or out-of-sample performance. Earlier notebooks retain their course structure so the progression remains visible.
