Spam Classifier

Built a spam classifier with an NLP preprocessing pipeline (tokenization, stop-word removal) feeding TF-IDF vectorization into a Multinomial Naive Bayes model.

Evaluated rigorously with precision, recall, F1, and a confusion matrix to control false positives on imbalanced spam-vs-ham data.

Surfaced the most informative tokens driving each prediction to make the model's behavior interpretable.