Dimensionality-reducing classifiers for Spanish winter maintenance of roadways

Diego M. Jiménez-Bravo, Javier Bajo, Esther Dopazo, Juan F. De Paz, Valderi Reis Quietinho Leithardt

Abstract: Public administrations try to maintain and guarantee safety as one of their priorities. Road networks are one of the environments where most incidents usually happen; therefore, it is very important to maintain the safety of drivers and passengers. This safety must be maintained even in adverse circumstances, such as winter weather conditions. Thus, chemical treatments are incorporated to be applied on roads when certain conditions are met. Determining which treatment to apply can be automated through the use of sensors and machine learning techniques. To this end, the article proposes a classifier model capable of predicting the most appropriate chemical treatment based on the environmental properties of the road and its surroundings. The achieved solution offers good results in terms of evaluation metrics.