Machine learning models to predict attraction waiting times at t+2h during an academic hackathon.

This project was developed during the M1 Introduction Week Hackathon, organized by Eleven Strategy.

The objective was to use historical data to predict waiting times two hours ahead (t+2h) for three attractions in a theme park, using supervised machine learning techniques.

• Raw data preprocessing and feature construction implemented in
  feature_engineering.py

Two approaches were explored:

• Linear Regression

  • Implemented in theme_park_linear.py
  • Achieved an RMSE of 9.54

• XGBoost Regressor

  • Implemented in theme_park_xgb.py
  • Improved performance with an RMSE of 8.81

Predictions were submitted as a CSV file with the following structure:

DATETIME | ENTITY_DESCRIPTION_SHORT | y_pred | KEY

This project focuses on:

• Feature engineering
• Model comparison
• Regression performance optimization under time constraints

It was completed in a hackathon setting, emphasizing rapid experimentation and iteration.