This document summarizes research on developing a spam filtering system for SMS messages on mobile devices. The researchers preprocessed a dataset of SMS messages, extracting features like word counts and frequencies. They applied logistic regression and gradient boosting models, achieving up to 98% accuracy. Significant features for detecting spam included words like "win", "urgent", and message length. Incorporating interaction terms further improved model performance. The system aims to minimize incorrectly flagging legitimate messages as spam.