This project uses data on the usage of public bike-sharing bikes together with weather data to predict the demand for bike-sharing bikes in each district of Munich.
Two different data sources have been used to set up this project:
- MVG: usage of public MVG bike-sharing bikes from 2019 to 2022
- Open-Meteo: historical weather data for Munich on an hourly basis from 2019 to 2022
The aim was, to obtain an hourly prediction of the number of bike rentals for each district in Munich, taking into account only the start time and location of rentals.
For this aim, an XGBoost Regressor is trained on the source data for each district. Predictions can be made through the API by specifying a date within two weeks in the future. This will trigger a query of Open-Meteo's weather forecasting API for the selected date which will be used for predicting the demand.
This project was conducted as 'final project' to finisch off the Le Wagon Data Science Bootcamp Munich. Many thanks to Alex, Archanaa, Jonathan and Jui for the great teamwork!