Nowadays, a large part of the world’s population lives in urban areas. The increase in the population living in cities makes it difficult to live in cities.Local administrators seek solutions to these problems in order to make cities more livable. In order to increase the quality of transportation, which is among the important issues for local governments, studies are carried out to establish smart transportation systems for public transportation companies.
Estimating travel times is an important tool in managing transportation operations. In addition, operations managers use these forecasts to coordinate future transportation operations. The public transport travel time prediction has been a well-researched topic; various researchers have predicted travel time using mathematical, statistical, and machine learning based models. Researchers also have compared the performance of these models.
It can be seen that studies in this field can be divided into statistical methods and machine learning methods. And number of researchers use one of these methods. But in this study, we use both statistical and machine learning techniques together. Also, the parameters of this study are different from the other studies in this literature. The effect of the weather on the travel time will be analyzed. In this respect it is thought that the study will contribute to the literature.
In the study, in addition to model tuning, sensitivity analyzes are also carried out according to the change of the k value in the KNN algorithm and the change of the threshold values used to determine outliers. we can say that, as can be seen from the analysis, the Catboost algorithm appears to be the algorithm that gives the best results in almost every situation.