Predict_Calorie-Expenditure

🥗 Calorie Prediction Model

This project aims to accurately predict the number of calories burned based on individual biometric and activity features using advanced ensemble machine learning techniques.

📌 Project Overview

In this project, we perform regression modeling to estimate the number of calories burned, using features like age, sex, height, weight, heart rate, body temperature, and exercise duration. We apply three powerful models — Random Forest, LightGBM, and CatBoost — and combine their predictions using a Stacking Regressor with XGBoost as the meta-model.

---

📁 Dataset

The dataset contains biometric and activity information of individuals, split into:

• train.csv: Contains both features and target (Calories)
• test.csv: Contains only features, used for generating predictions

🎯 Features

Column	Description
id	Unique identifier
Sex	Gender of the individual
Age	Age in years
Height	Height in centimeters
Weight	Weight in kilograms
Duration	Exercise duration (mins)
Heart_Rate	Avg heart rate during exercise
Body_Temp	Body temperature (°F)
Calories	🔺 Target variable

---

⚙️ Workflow

✅ 1. Exploratory Data Analysis (EDA)

• Univariate and bivariate analysis
• Outlier Detection
• Distribution checks for numeric features
• Correlation analysis

🧹 2. Data Cleaning & Preprocessing

• Encoding categorical features (Sex)
• Scaling numerical features using RobustScaler
• Train-Test splitting

🔧 3. Feature Engineering

• Interaction terms based on domain logic
• Handled skewness if required

🧠 4. Model Building

Base Models

• RandomForestRegressor
• LGBMRegressor
• CatBoostRegressor

Meta Model

• XGBRegressor — combines predictions from base models in a stacked ensemble

Final Model

• StackingRegressor with passthrough enabled

---

📈 Final Evaluation

Metric	Score
RMSE	3.5823
MAE	2.1463
R² Score	0.9967

📂 A submission file was generated using predictions on the test dataset.

---

🛠️ Installation & Setup

1. Clone the repository

git clone https://github.com/rohankharche34/calorie-prediction-model.git
cd calorie-prediction-model

2. Create and activate a virtual environment (optional but recommended)

python -m venv venv
source venv/bin/activate  # on Windows use `venv\Scripts\activate`

3. Install dependencies

pip install -r requirements.txt