Lost Time Prediction Model of Left-turn Signal Based on Random Forest Regression
Left-turn signal loss time is an important parameter for determining the signal control cycle.In order to accurately predict the left-turn signal loss time and optimize the signal control scheme,the left-turn signal loss time was analyzed in terms of influencing factors from the overall environment of the intersection,driver behavior,and traffic flow characteristics.Qualitative and quantitative methods were used to collect data from different intersections in Honggutan District,Nanchang City.Three different left-turn signal loss time prediction models of random forest regression,multiple linear regression and generalized linear mixed were constructed and analyzed by analyzing the influence of each influencing factor on the left-turn signal loss time.The results show that the predict accuracy of the left-turn signal loss time by random forest regression model was higher than that by the traditional linear model,with a coefficient of determination of 0.812 9;The reaction time of the first vehicle was the most important factor influ-encing the prediction of the left-turn signal loss time(with an importance score of 14.5),followed by the way of or-ganizing the left-turn,the time of the red light,and the distance of the opposite stop line,with importance scores of 2.88,2.84,and 2.23 respectively;The left-turn signal loss time was strongly influenced by start-up loss time.
signal-controlled intersectionleft-turn traffic flowrandom forest regression modelleft-turn signal loss time
万方数据
维普
