🛒 Shopper Intent Prediction System

An end-to-end Machine Learning Engineering project that predicts whether an online shopping session will result in a purchase. The project demonstrates the complete ML lifecycle, including data preprocessing, feature engineering, model training, benchmarking, production model selection, REST API deployment, monitoring, and automated testing.

---

🚀 Project Highlights

• End-to-end machine learning pipeline
• Automated preprocessing and feature engineering
• Multiple classification models
  • Logistic Regression
  • Decision Tree
  • Random Forest
  • Artificial Neural Network (ANN)
• Weighted production model selection
• FastAPI prediction service
• Data quality and feature drift monitoring
• Comprehensive unit testing using Pytest
• Modular and configuration-driven architecture

---

🏗️ System Architecture

---

📂 Repository Structure

.
├── api/                    # FastAPI application
├── configs/                # Training configuration
├── data/
│   ├── raw/
│   └── processed/
├── docs/                   # Project documentation
├── models/                 # Trained models and deployment artifacts
├── reports/                # Monitoring reports
├── src/
│   └── shopper_intent/     # ML pipeline implementation
├── tests/                  # Unit tests
├── README.md
└── pyproject.toml

---

⚙️ Quick Start

1. Clone the repository

git clone <repository-url>
cd shopper-intent-prediction

2. Install dependencies

pip install -r requirements.txt

or

poetry install

3. Train the models

python -m src.shopper_intent.train

4. Start the API

uvicorn api.main:app --reload

Open:

http://127.0.0.1:8000/docs

---

🌐 API Example

Request

POST /predict

{
    "Administrative": 2,
    "Administrative_Duration": 45.0,
    "Informational": 1,
    "Informational_Duration": 12.5,
    "ProductRelated": 18,
    "ProductRelated_Duration": 620.3,
    "BounceRates": 0.02,
    "ExitRates": 0.05,
    "PageValues": 18.7,
    "SpecialDay": 0.0,
    "Month": "Nov",
    "OperatingSystems": 2,
    "Browser": 1,
    "Region": 3,
    "TrafficType": 2,
    "VisitorType": "Returning_Visitor",
    "Weekend": false
}

Response

{
    "prediction": 1,
    "probability": 0.9437,
    "model": "logistic_regression",
    "model_version": "1.0.0"
}

---

🧪 Running Tests

Execute the unit test suite using:

pytest

Verbose output:

pytest -v

---

📚 Documentation

Detailed technical documentation is available in the docs/ directory.

Document	Description
📖 01. System Design	System architecture, component design, project structure, data flow, and design decisions.
🤖 02. Model Development	Dataset, preprocessing, feature engineering, model development, evaluation, benchmarking, production model selection, and model artifacts.
📊 03. Monitoring Strategy	Data quality monitoring, feature drift detection, retraining recommendation, monitoring reports, and limitations.
🌐 04. API Documentation	REST API endpoints, request and response schemas, validation, and deployment instructions.
🧪 05. Testing	Testing strategy, unit test coverage, execution instructions, and testing limitations.

---

📖 Documentation Roadmap

README
│
├── 01. System Design
│
├── 02. Model Development
│
├── 03. Monitoring Strategy
│
├── 04. API Documentation
│
└── 05. Testing

The README provides a high-level overview of the project, while the docs/ directory contains detailed technical documentation for each stage of the machine learning engineering workflow.

---

📈 Technologies

Category	Technologies
Programming Language	Python
Machine Learning	Scikit-learn, TensorFlow/Keras
Data Processing	Pandas, NumPy
API	FastAPI
Testing	Pytest
Configuration	YAML
Model Serialization	Joblib

---

📄 License

This project was developed for educational purposes as part of a Machine Learning Engineering assignment.