To predict the Contributing factor for the top one hundred collisions. BayesNet, Naïve Bayes, J48, and KNN classification algorithms were used. Figure 18 shows the accuracy, MAE, and RMSE, for each algorithm. Naïve Bayes had the best accuracy level of 59%, with the lowest MAE. Concluding Naïve Bayes is the most fitting algorithm to predict a collision’s contributing factor.

Assumptions