Deep Learning incident prediction
Deep learning prediction algorithm
We have developed a platform to ingest road data, real-time traffic data and environmental data, such as weather, to predict the risk of incidents 15-30 minutes in the future, in real-time. The models detect traffic patterns and other information signals that – on a specific location – increase the accident probability. We predict accident probability on road segments of various lengths for large scale networks at regional and national levels.
- The algorithm has been trained on 2 years of traffic and incident data for the whole National Road Network in the Netherlands.
The risk is calculated for 3,128 segments with a total length of 6,890 km.