Our goal in AI-CENTIVE is to help Austria achieve important climate goals by incentivizing citizens to use more sustainable forms of transportation. To do this, we need to understand current mobility preferences, what alternatives are possible, and which incentives would motivate people to switch to those alternatives.
Ten months into our project, the first achievements are in place to reach the project’s goal. A fundamental building block is an AI model trained on past mobility behaviour to predict what future mobility behaviour will look like – without our incentives. To accommodate all the contexts that can influence how and when we travel, it was necessary to gather data from multiple sources – not only the actual mobility behaviour of Austrian citizens, but also data related to weather, news feeds, and knowledge extraction from the Web to identify possibly relevant events.
We have evaluated models using the latest state-of-the-art approaches in deep learning and predictive analytics, in order to select which model can be consistently more accurate with our data and also to identify how each of the input features (e.g. weather, news, events) contributes to the accuracy.
The next steps will be to take the best performing model, make predictions about future mobility in Austria with it, and train it to learn how mobility choices change with incentives based on user behaviour with the “ummadum” app.
An AI model for mobility choice prediction is an important starting point for our work. Without it, we cannot know how incentives would change citizens’ mobility decisions, as we cannot predict what those decisions would otherwise be.
In the coming months, we will be reporting the insights we gain from this model on current Austrian mobility trends as well as on the Incentivisation Pilot, which will give us the valuable data we need on which incentives influence which mobility decisions to teach our AI about incentives.
Are you keen on learning more about the AI implementation in sustainable mobility projects? Contact us at office@dataintelligence.at. Stay informed about crucial updates on this subject and subscribe to the AI-CENTIVE newsletter.
Author: Asst.-Prof. Dr. Lyndon Nixon
