Traffic Incident Detection Method Based on Machine Learning

Abstract

This paper aims to scale back the Traffic incidents entitled “TRAFFIC INCIDENT DETECTION technique supported MACHINE LEARNING” . Timely and actual detection of traffic incidents will effectively cut back personal casualties and property losses, and improve the capability of macro-control and scientific decision-making of traffic. The unbalance of traffic incident knowledge features a nice impact on the detection impact. Therefore, a traffic incident detection technique supported machine learning (FA-WRF) is intended. Through the analysis of the amendment rule of traffic flow framework to make the initial incident variable. The correlational analysis (FA) technique is employed to scale back the extent of the initial incident variables. victimization Bootstrap improved algorithmic rule to fate the information extraction normal of the coaching set. The Medical counseling Committee constant worth is calculated for the classification impact of the choice tree when coaching, and is allotted to every tree as a weight worth, therefore on make sure that the trees with higher classification capability have additional ballot power within the ballot method, so improve the classification performance of the random forest (RF) algorithmic rule for unbalanced knowledge.