A machine learning-based binary classifier that detects spam emails using text preprocessing and statistical features. Built using Scikit-learn, this model analyzes email contents and filters spam using classical NLP + ML techniques.
🎯 A foundational classification project demonstrating end-to-end spam detection using TF-IDF, logistic regression, and model evaluation on real-world datasets.
Email spam remains one of the most prevalent digital nuisances. This project trains a supervised learning model to classify whether an email is spam or not based on its textual content. The pipeline includes preprocessing, vectorization, training, and performance evaluation.
-
Data Preprocessing
- Lowercasing, punctuation removal, stopword filtering
- Lemmatization using
nltk
-
Vectorization
- TF-IDF applied on cleaned email text
-
Modeling
- Logistic Regression (primary classifier)
- Baseline: Naive Bayes
- Evaluation: Accuracy, confusion matrix, ROC AUC
-
Result
- Achieved 97.9%+ accuracy and robust spam detection performance.
| Metric | Value |
|---|---|
| Accuracy | 97.98% |
| Precision | 98.43% |
| Recall | 97.65% |
| AUC Score | 0.991 |