A minimalistic machine learning app to classify email or message text as Spam or Ham, built using Streamlit, scikit-learn, and a trained model from the UCI SMS Spam Collection Dataset.
This project uses the UCI SMS Spam Collection Dataset, a well-known benchmark dataset consisting of 5,574 labeled SMS messages, divided into:
- ham → legitimate (non-spam) messages
- spam → unsolicited, promotional, or fraudulent messages
The machine learning model analyzes input text and predicts whether the message is Spam or Ham based on:
- Keyword frequency
- TF-IDF vector patterns
- Statistical patterns commonly seen in spam messages
This enables fast text screening for:
- Email filtering
- SMS moderation
- Basic content risk assessment
The model is trained using TF-IDF Vectorization combined with a Machine Learning classifier (Support Vector Machine).