Improved Accuracy of Calculation of Vehicle Crash Severity in Highways using Random Forest over Multi-Layer Perceptron Algorithm

Abstract

Aim: To improve the accuracy rate of vehicle crash severity in highways using Random forest over MultiLayer Perceptron. Materials and Methods: Random forest and MultiLayer Perceptron with sample size of (N=10) is executed with varying training and testing splits for calculating the accuracy for accident crash severity with g power as 75%, threshold 0.000 and confidence interval 95%. The performance of the classifiers are evaluated based on their accuracy rate using accident severity dataset. Results: The accuracy for calculating accident crash severity in Random Forest(91%) and MultiLayer Perceptron (83%) is obtained(p<0.005). Conclusion: Prediction of accident crash severity using Novel Random Forest (RF) algorithm appears to be significantly better than MultiLayer Perceptron (MLP) with improved accuracy.