Bikeshare Wizard was a tool developed for my thesis during my Masters of Information and Data Science at Berkeley. I worked with some awesome teammates (see below) to develop a tool to improve the user experience for bikeshare users (think Citibike in NYC)

Our team was frustrated with the current status of bikeshare systems. We had all experienced the disappointment of getting to a station only to find that no bikes were there, or trying to return a bike to a station only to find that no docks were available.

We wanted to make a tool that could predict the number or bikes and docks available for a given station, at a set point in time.

We used deep learning models (primarily DeepAR from Amazon) and data science and machine learning techniques to build a tool that was capable of doing this

Our project was picked as one of the top thesis projects, and was nominated for the Hal R Varian Award for best thesis

Bikeshare Wizard is no longer online, but click below to download our presentation deck

Deck here