Spam Email Filter

In this project, I developed an email spam filter using classification techniques. The goal was to build accurate models to distinguish between legitimate and spam emails, improving user experience and mail security.

Highlights

• Bayesian Baseline: Implemented a probabilistic model as a benchmark for classification accuracy.
• SVM Model: Designed and tuned an SVM-based classifier to better separate spam from ham emails.
• Data Preprocessing: Cleaned and tokenized email text, extracted features, and structured training data.
• Model Development: Built and tuned both models using real-world datasets, improving with feature engineering and validation.
• Evaluation & Comparison: Measured performance using precision, recall, F1-score, and accuracy to compare the models’ effectiveness.
• Real-world Applicability: Demonstrated relevance for mail services seeking to reduce inbox spam automatically.

Outcomes

The project resulted in a functional spam classification pipeline using both probabilistic and margin-based models. The SVM model showed significant improvements over the baseline.

Attachments