Crystalball

Inspiration

Giving visibility to unexpectedly crowded travel routes

What it does

Provides a heatmap of Berlin for crowdedness of stations based on machine-learning predictions

How I built it

Basic random forest model. Using BVG data.

Challenges I ran into

Only portion of all trams provided data

Accomplishments that I'm proud of

Integrated the predictions on BVG data into a frontend

What I learned

Using Leaflet for visualization

What's next for Crystalball

Online improvement of machine-learning by adding new data

Built With

• java
• leaflet.js
• pandas
• scikit-learn