🔥 TrackFit(Personal Fitness Tracker)

Welcome to the Personal Fitness Tracker! This Streamlit-based web application allows users to predict the calories burned during exercise based on their personal details. The app leverages machine learning models to provide accurate predictions and helps users track their fitness journey over time.

🖥️ Screenshots

Home Page

Prediction Result

---

🚀 Features

• Input BMI directly or calculate it from height and weight.
• Predict calories burned using SVM, Logistic Regression, or Random Forest models.
• Automatically selects the best model based on user parameters.
• Generates a unique User ID for tracking predictions.
• Stores and displays past predictions.
• Allows downloading past predictions as CSV.
• Provides historical data visualization.
• Displays feature importance for Random Forest models.

---

📦 Installation Guide

1️⃣ Install Required Dependencies

pip install -r requirements.txt

2️⃣ Ensure Model Files Are in Place

Make sure you have the pre-trained model files in the models/ directory:

models/
├── svm_model.pkl
├── logistic_regression_model.pkl
└── random_forest_model.pkl

If these files are missing, retrain and save models using joblib before running the app.

3️⃣ Run the Streamlit App

streamlit run app.py

---

🛠 Technologies Used

• Python
• Streamlit (for UI and interactivity)
• Pandas (for data processing)
• Scikit-Learn (for machine learning models)
• Joblib (for model storage)
• Matplotlib & Seaborn (for visualizations)

---

📊 Data & Model Details

• The app supports three machine learning models: Support Vector Machine (SVM), Logistic Regression, and Random Forest.
• Based on BMI, age, and workout duration, the app selects the most appropriate model.
• Predictions are stored in data/history.csv for tracking and analysis.

---

🏋️‍♂️ How It Works

1. Enter personal details such as BMI, age, gender, workout duration, heart rate, and body temperature.
2. App selects the best model automatically.
3. Click "Start Prediction" to get the estimated calories burned.
4. Predictions are saved, allowing users to track progress.
5. Download historical predictions as CSV for further analysis.

---

📄 License

This project is licensed under the MIT License.