Email Classifier with PII masking

This project demonstrates how to build an email classifier system to :

• Mask personally identifiable information (PII) like name, email, phone number, card number from emails.
• Classify emails into categories like 'Request', 'Incident', 'Change', 'Problem'.

---

Tech Stack

• Language: Python 3
• Framework: FastAPI
• ML: Scikit-learn (Multinomial Naive Bayes)
• Text Preprocessing: Regex, SpaCy (for NER)
• Model Saving: Joblib
• Deployment Ready: Yes, via Uvicorn server

---

Features

• Regex + Spacy powered PII masking
• Trained Multinomial Naive Bayes classifier (TF-IDF)
• REST API built with FastAPI
• Strict JSON format output
• Ready for deployment and production usage

---

Model Performance

• Accuracy: 71.5%
• Classifier: Multinomial Naive Bayes with TF-IDF vectorization
• Most Confident Class: Request (F1 = 0.88)

Example Request & Response

Request

Response

Project Structure

email-classifier
| |--data
|--combined_emails_with_natural_pii.csv |--saved_model
|--model.pkl |--tfidf.pkl |--api.py |--model.py |--utils.py |--README.md |--requirements.txt

How to Run Locally

1. Clone the Repository

git clone < repo-link >
cd email_classifier_project

2. Create Virtual Environment (Optional but recommended)

python -m venv email_classifier
source email_classifier/bin/activate  # Mac/Linux
email_classifier\Scripts\activate     # Windows

3. Install Requirements

pip install -r requirements.txt
python -m spacy download en_core_web_sm

4. Start the FastAPI Server

uvicorn api:app --reload

Test the API (Swagger UI)

- Open your browser and go to: http://127.0.0.1:8000/docs
- Use the /classify-email endpoint to test!

Future Improvements

• Use n-gram features for better phrase detection
• Try SVM or Logistic Regression
• Experiment with contextual models like BERT
• Add logging and exception handling to the API
• Containerize the app using Docker