Accident Risk Detection in Urban Trees using Machine Learning and Fuzzy Logic

Knowing the state of trees and their associated risks contribute to the care of the population. Machine Learning, through supervised learning, has demonstrated its effectiveness in various areas of knowledge. The risk of accidents can be predicted by having different tree data, including height, species, condition, presence of pests, the area where it is planted, climatic events, and age. This work proposes a platform to register trees and predict their risk. The solution considers integrating technology and applications for those in charge of maintenance and changes in current procedures. The risk prediction process is carried out through a fuzzification process that contributes to the responsible entities' decision-making. Preliminary results of this research are presented, and the capacity of the developed software architecture is demonstrated, where the scalability of the prediction algorithm stands out.