Research On The Task Assignment Of Spatial Crowdsourcing Taxi Based On Machine Learning

In recent years,with the rapid development of "Internet plus initiative" economy,the connection between the Internet and traditional industries is becoming closer and closer,which also brings new impetus to the industry.In terms of taxi trip,the development of online car-hailing is particularly rapid.There are many online carhailing platforms in the market.With its convenient and high-quality passenger experience,it is more and more accepted by the general public.In order to further improve the matching between online booking and passengers,alleviate the contradiction between supply and demand;shorten the waiting time for passengers,and increase the success rate of taxis in hot areas and peak hours.This paper studies the allocation and scheduling algorithm of the online car-hailing platforms.Through real-time traffic,weather and other data in various areas of the city,the dynamic prediction of the short-term supply and demand of online car-hailing in the city can be realized.The short-term forecast results can be used to sense the hot spots and shortages of taxis in advance,Help the platform to achieve effective scheduling of online car-hailing,improve the success rate of passengers and reduce waiting time;for drivers can increase the number of daily orders,and improve the overall rate of online car-hailing.In response to the above problems,the supply-demand relationship between online car-hailing and short-term passenger travel is first studied,and the current machine learning algorithms for short-term demand forecasting for online car-hailing.Using more than 6.8 million taxi data,the impact of various types of data in the city such as weather conditions,temperature,and traffic congestion on passenger travel is analyzed.In order to obtain a short-term passenger travel demand forecasting model and a shortterm taxi supply-demand gap forecasting model,through the feature extraction and construction of the data set,the gradient prediction decision tree algorithm,extreme gradient improvement algorithm,and random forest algorithm are used to train the prediction model.The regression model performs model fusion,which further improves the accuracy of the model prediction.According to the actual operation data test,the good effect of the short-term demand forecasting model of online car-hailing in this paper is verified.Secondly,in order to address the problem of crowdsourcing tasks in the space of online car-hailing,based on the short-term demand forecast of online carhailing,we can detect the shortage of vehicles in the short-term hotspots in the future and the relationship between supply and shortage in the area.Scheduling has effectively improved the success rate of passengers during peak hours in hot spots.Finally,the experimental results show the effectiveness of the crowdsourcing task allocation algorithm proposed in this paper.