Calories Burnt Prediction using XGBoost

Overview

This project predicts the number of calories burnt during physical activity using the XGBoost machine learning algorithm. It processes a dataset containing various biometric and activity-related features to train a regression model.

Dataset

The dataset includes the following CSV files:

• exercise.csv: Contains features such as heart rate, duration, and type of physical activity.
• calories.csv: Contains the corresponding number of calories burnt for each exercise session.

Features Used

Some of the features used in model training include:

• Age
• Gender
• Height
• Weight
• Duration
• Heart Rate
• Body Temperature

Dependencies

Ensure the following Python packages are installed:

• Python 3.x
• pandas
• numpy
• xgboost
• scikit-learn
• streamlit (for the web interface)

To install all dependencies at once, use:

pip install -r requirements.txt

Additional Setup for macOS

If you're on macOS and encounter errors related to protobuf, install it using Homebrew:

brew install protobuf

Usage

Option 1: Run in Google Colab

Use Calories_Burnt_Prediction.ipynb for the complete pipeline:

• Data loading
• Preprocessing
• Model training
• Evaluation

Option 2: Run Web App Locally Using Streamlit

A web interface is provided via Streamlit for user input and calorie prediction.

1. Activate your Python environment.
2. Install dependencies:

   pip install -r requirements.txt

3. Run the Streamlit app:

   streamlit run calories_app.py

This launches a browser-based UI where you can enter values and view predicted calories burnt.

Files

• README.md: This file.
• Calories_Burnt_Prediction.ipynb: Google Colab notebook containing preprocessing, training, and evaluation code.
• calories_app.py: Streamlit script for launching the web-based prediction interface.
• requirements.txt: List of required Python packages.