Taxi Order Forecasting Based on Weighted Average Fusion Model
In order to improve the efficiency of taxi operation and realise reasonable dispatching of taxis,and to address the problems of low accuracy and single consideration of factors in the current single model for traffic prediction problems,this paper adds four meteorological influencing factors on the basis of GPS data of taxis in Chongqing to study the law of taxi orders in the hot-spot areas of ridesharing.A weighted average fusion model with LSTM,ARIMA and CNN as sub-models is used to predict the num-ber of taxi orders in the hotspot areas to improve the prediction accuracy.The results show that the error normalised weighted aver-age fusion model achieves better prediction results compared to both other fusion methods and single prediction models,and is more suitable for forecasting taxi demand in hotspot areas.
国家科技期刊平台
NSTL
万方数据
