Dreiländereck

• 

  Model performance on unseen test data

• 

  Model performance shown on one saturday

• 

  Model performance shown on one tuesday

Inspiration

Help people have a better experience using public transportation and acquire more people for public transport. This would lead to a positive impact on congestion, environment and air pollution due to fewer vehicles.

What it does

Use historic traffic volume and weather data to predict future traffic appearance of traffic in Ingolstadt. From these predictions make proposals when the best times to commute are.

How we built it

Pulling the data from two different APIs. Merging it to a complete usable data set. Training a regression model on the pre-processed data. Evaluating it's performance on unseen data.

Challenges we ran into

Getting the data we wanted in the time we had, as well as in the quality and quantity we expected it to be.

Accomplishments that we're proud of

Finishing on time. Having measurable results. Having fun in a completely new environment.

What we learned

It is important to have a precise idea of what to accomplish in the given time. How important the quality of the data is to get good results with machine learning models.

What are the next steps planned for the project

Merging it with passenger data, delay data of busses. Increasing the data used to the whole city of Ingolstadt. Further increasing the quality of the prediction model.

Built With

• open-meteo-weather-api
• python
• regression-models
• scikit-learn
• sensorthingsapi