Prediction of the Delay Time of Public Transportation Using Machine Learning

Reliable urban transport is one of the most important elements of an efficiently operating city. However, public transport often disappoints travelers due to a lack of comfort, unpredictability, and frequent delays. For this reason, many people still choose cars, which negatively influence the environment and, as a result, the quality of life in a city. Public transport has many advantages, but undoubtedly, one of its most significant disadvantages is the problem with punctuality. The aim of the research is to develop a machine learning based method for predicting tram delays in Wrocław, Poland. We develop Random Forest Regressor, XGBRegressor, and deep neural network models.Our contribution is related to the integration of the following sets of data: location of public transport vehicles on the route, public transport timetable, weather, and air quality for improving the prediction performance. Such an approach has not yet been developed. The most effective solution was a model based on a random forest regressor. The selected model showed very good results regarding the level of data explanation and low prediction error. Data analysis showed that the average delay time exceeds 1.5 min.