Email Spam Detector
NLP-Powered Classification with 100% Accuracy
Project Overview
An NLP-powered email spam detector that achieves perfect classification using machine learning. Trained on 320 emails, the system uses TF-IDF vectorization and multiple algorithms to identify spam patterns with 100% accuracy.
Built using Python, scikit-learn, and natural language processing techniques, this project demonstrates text preprocessing, feature extraction, and multi-model comparison.
Top Spam Word Indicators
Top 15 words that appear most frequently in spam emails. "Free", "cash", and "prizes" are the strongest indicators of spam content.
Model Performance Comparison
All three models (Naive Bayes, Logistic Regression, Random Forest) achieved perfect 100% accuracy on the test set.
Confusion Matrix
Perfect classification with zero false positives or false negatives on 64 test emails.
Using the Model
The trained model is saved and ready to classify new emails. The spam detector can be used to automatically filter incoming messages.
Quick Start:
Example Classifications:
Key Features:
- Real-time email classification using saved model (
spam_detector.pkl) - TF-IDF vectorization for text feature extraction
- Confidence scores for each prediction
- Handles text preprocessing automatically (lowercase, special character removal)
- Trained on balanced dataset (169 spam, 151 ham)