Training-ML 🚀

🎯 Objective

This project implements a complete Machine Learning pipeline for binary classification (income prediction).

It demonstrates how to:

• preprocess data
• train a model
• evaluate performance
• visualize results
• save and reuse a model
• expose predictions via a REST API

---

⚙️ Tech Stack

• Python
• Pandas
• Scikit-learn
• NumPy
• Matplotlib
• Seaborn
• Flask

---

📊 Project Overview

This project follows a standard Machine Learning workflow:

1. Load dataset
2. Clean and preprocess data
3. Train a model (Random Forest)
4. Evaluate performance
5. Visualize results
6. Save trained model (model.pkl)
7. Serve predictions via API

---

📁 Project Structure

training/
│
├── data/
│   └── dataset.csv
│
├── src/
│   ├── train.py
│   ├── predict.py
│   ├── preprocess.py
│   ├── metrics.py
│   ├── visualization.py
│
├── models/
│   └── model.pkl
│
├── flask/
│   └── api.py
│
├── app.py
├── requirements.txt
└── README.md

🚀 Installation

Clone the repository and install dependencies:

pip install -r requirements.txt

---

▶️ Run the Training Pipeline

python app.py

This will:

• load and clean data
• train the model
• evaluate performance
• generate visualizations
• save the model in /models/model.pkl

---

🌐 Run the API

python flask/api.py

API will run on:

http://127.0.0.1:5000

---

📡 API Usage

Endpoint

GET /
GET /health
GET /visualization
POST /predict

---

Example Request

{
  "features": [1, 50, 35, 50000, 0, 1, 3, 2, 12, 100, 5]
}

---

Example Response

{
  "prediction": 0,
    "probabilities": {
        "0": 0.7916445304104879,
        "1": 0.20835546958951212
    }
}

---

Error Example

{
  "error": "Missing features"
}

---

📈 Model Output

During training, the model prints:

• Accuracy score
• Confusion matrix
• Classification report

It also displays:

• Confusion matrix visualization

---

🧠 Model Details

• Algorithm: Random Forest Classifier
• Task: Binary classification
• Target: PINCP (income threshold classification)
• Train/Test split: 80/20
• Fixed random state for reproducibility

---

💡 Key Features

• Clean and modular code structure
• Reusable and scalable ML pipeline
• Model persistence with .pkl
• API for real-time predictions
• Data visualization included

---

🔧 Future Improvements

• Add cross-validation
• Improve feature engineering
• Add model comparison (Logistic Regression, XGBoost)
• Deploy API (Render, AWS, etc.)
• Build a frontend interface

---

📌 Notes

• Ensure your dataset contains the PINCP column
• Input features must match training data format
• API expects numerical input only

---

👨‍💻 Author

Machine Learning beginner project focused on building a complete and structured ML pipeline.

---