02 · Machine Learning · Tourism · 2025
TourSight.
A forecasting tool that combines historical visitor data, weather and temporal patterns to predict crowd levels at tourist attractions.
Overview
TourSight was built in a 24-hour sprint at the 2025 Tourism Technology Hackathon and received the award for Best Advanced Technology.
Challenge
Raw sensor counts show what is happening now, but tourism teams need reliable forecasts to plan staffing, visitor flows and operations in advance.
Approach
We engineered more than 50 temporal and weather features, trained a LightGBM model and surfaced 24-hour forecasts in 15-minute intervals through a React dashboard with D3.js visualizations.
Outcome
The prototype achieved an average forecast error of roughly six visitors and demonstrated how existing data can become an actionable planning tool.