Phishing Email Classifier

Overview

I built this project to classify phishing emails based on the text of the email using machine learning. I trained a logistic regression model.

Dataset

• The training dataset contains 8,348 labeled examples
• The test dataset contains 1,000 unlabeled examples

Data Format:

Each email has the following columns:

• id
• subject line
• email (body text).
• values (a label where 1 is a phishing email and 0 is a legitimate email)

Project Components

1. EDA, Feature Engineering

   • Extract features from email text, visualize.

2. Modeling

   • Apply logistic regression for classification.
   • Use cross-validation to evaluate model performance and prevent overfitting.

3. Evaluation

   • acuraccy, precision, recall, F-1 score.

4. Test on unlabeled data

Credits

The dataset is from a Data Science class at UC Berkeley.