Comparative Analysis using K - Nearest Neighbour with Artificial Neural Network to Improve Accuracy for Predicting Road Accidents

Abstract

The purpose of this study is to use machine learning approaches to improve the accuracy of modern road accident prediction systems like the K-Nearest Neighbour Algorithm and Artificial Neural Networks Algorithm. Materials and techniques used include the K-Nearest Neighbour technique and the Artificial Neural Networks algorithm with sample size N=10, iterated 20 times in parallel to test the accuracy of forecasting road accidents. p0.05 indicates the significance of the K-Nearest Neighbour method. When comparing the results of the two algorithms, it is discovered that the K-Nearest Neighbour approach (81.22%) outperforms the Artificial Neural Networks algorithm (69.22%) in terms of accuracy in forecasting road accidents.