PROJECT // MACHINE LEARNING APPLICATION
SpamShield
An email spam classification application using TF-IDF features with Naive Bayes and Logistic Regression models.
Problem
Email text must be transformed into useful numerical features before a classification model can distinguish spam from legitimate messages.
Solution
SpamShield preprocesses email text, applies TF-IDF vectorization, and compares predictions from two classical machine-learning models through a simple web interface.
Architecture
A browser interface sends email text to a Python-backed classification workflow. Preprocessing and vectorization feed Naive Bayes and Logistic Regression models, whose predictions and confidence are returned to the user.
Technologies and features
- / Real-time email classification
- / Naive Bayes and Logistic Regression model comparison
- / Confidence score display
- / Text preprocessing and TF-IDF features
- / Responsive browser interface
Engineering decisions
Using two lightweight, interpretable classifiers makes model behavior easier to compare while keeping prediction fast for an interactive demo.
Outcome
The project connects an applied machine-learning pipeline with a usable web interface and a published research reference.