Datta

Predicting bike-share usage patterns with machine learning

This thesis looks at how machine learning algorithms might be used to predict bike-share traffic. We determine the accuracies of estimators such as decision trees, random forests and boosted decision trees. The effect of factors such as weather, geographic location, time of day, day of week etc on the number of bikes at a bike-share station are also investigated. Finally, we outline how a web-based prediction system that uses the estimators mentioned in this thesis could look like.

The vision for this thesis is a prediction software that lets a user ask questions like this (image courtesy of istockphoto):

The thesis in it's entirety can be found here and the software:

The presentation can be found here.

Last modified Dec. 1, 2014 12:49 PM by arnabkd@uio.no
Last modified May 31, 2023 6:12 AM by root@localhost
Last modified May 31, 2023 6:12 AM by root@localhost
Last modified Dec. 4, 2014 9:50 PM by arnabkd@uio.no
Last modified May 31, 2023 6:12 AM by root@localhost