A Machine Learning solution developed for the Kaggle Playground Series competition.
The project predicts calorie expenditure during exercise using physiological measurements, engineered health metrics, and ensemble regression models.
- Kaggle Playground Series - Season 5 Episode 5
- Problem Type: Regression
- Final Rank: 2543 / 4316 Teams
Accurately estimating calorie expenditure is important for fitness tracking and health monitoring.
This project applies feature engineering techniques based on exercise science and trains machine learning models to predict burned calories.
Predict the number of calories burned during physical activity.
The dataset includes:
- Age
- Height
- Weight
- Duration
- Heart Rate
- Body Temperature
- Gender
Several health-related features were engineered, including:
- Calories Estimated
- Body Mass Index (BMI)
- Basal Metabolic Rate (BMR)
- Total Daily Energy Expenditure (TDEE)
- Lean Body Mass (LBM)
- Body Fat Percentage
- Activity Factor
- BMI Categories
- Fat Categories
These engineered features significantly improved the predictive performance.
Models used:
- Gradient Boosting Regressor
- XGBoost Regressor
Cross-validation was used during model evaluation.
Regression models were evaluated using:
- RMSLE (Root Mean Squared Logarithmic Error)
The final predictions were exported for Kaggle submission.
- Python
- Pandas
- NumPy
- Scikit-learn
- XGBoost
- Matplotlib
- Seaborn
- Jupyter Notebook
notebooks/
screenshots/
README.md
requirements.txt
This solution was fully developed by me.
Responsibilities included:
- Data preprocessing
- Feature engineering
- Health metric calculation
- Model training
- Cross-validation
- Prediction generation
- Kaggle submission creation
pip install -r requirements.txtRun the notebook to reproduce the results.
Karim Maher
Data Science | Machine Learning | Flutter | Digital Marketing