Email Classifier

A machine learning tool that automatically classifies emails into categories such as "notifications" and "correspondence" based on their content and metadata.

Features

• Training: Train the classifier using your own categorized emails
• Evaluation: Evaluate the classifier's performance with test data
• Classification: Classify new emails automatically
• Model Backups: Automatically creates backups of previous models when training new ones

Installation

Prerequisites

• Python 3.6 or higher
• pip (Python package installer)

Setup

1. Clone the repository:

   git clone https://github.com/cpknight/email-classifier.git
   cd email-classifier

2. Install the required dependencies:

   pip install pandas scikit-learn nltk

Usage

The email classifier script supports three main commands: train, evaluate, and classify.

Training the classifier

./email_classifier.py train --dir <training_directory> --output <model_file>

Example:

./email_classifier.py train --dir ./emails --output model.pkl

By default, the script expects your training data to be organized in two subdirectories:

• notifications/: Contains notification-type emails
• correspondence/: Contains correspondence-type emails

When training a new model, if a model already exists at the specified output path, a backup will be created with the original creation timestamp appended to the filename.

Evaluating the classifier

./email_classifier.py evaluate --model <model_file> --dir <test_directory>

Example:

./email_classifier.py evaluate --model model.pkl --dir ./test_emails

Classifying emails

./email_classifier.py classify --model <model_file> --email <email_file>

Example:

./email_classifier.py classify --model model.pkl --email ./new_email.eml

Data Organization

For training and evaluation, the script expects emails to be organized in directories by category:

training_data/
├── correspondence/
│   ├── email1.eml
│   ├── email2.eml
│   └── ...
└── notifications/
    ├── email1.eml
    ├── email2.eml
    └── ...

Each .eml file should be a standard email file with headers and content.

Model Backup Feature

When training a new model, if a model already exists at the specified location, the script automatically creates a backup of the existing model before overwriting it. The backup filename includes the original creation timestamp in the format:

model.pkl.YYYYMMDD_HHMMSS.bak

This feature ensures you never lose previous model versions and can track model evolution over time.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Author

Created by cpknight ... I can't take credit. Claude 3.7 sonnet by way of Warp wrote it - this is another AI-generated project!