Spam Classifier

A very Spam Classifer built using TF-IDF Vectorizer and tested on multiple split parameters.

• Two models are compared for performance: Logistic Regression and Random Forest Classifier.
• Both models are properly evaluated with classification reports to show performance based on precision/recall.
• Classification capabilities are also analyzed using feature importances from both models, showing how agressive and conservative each model is in classifying certain words as spam.
• The models have been run twice, both with and without stop words.

Logistic Regression Results:

Class	Precision	Recall	F1-Score	Support
ham	0.97	1.00	0.99	958
spam	0.98	0.83	0.90	157

TF-IDF + RandomForestClassifier:

Class	Precision	Recall	F1-Score	Support
ham	0.98	1.00	0.99	958
spam	1.00	0.85	0.92	157