Research On Hotspot Recognition And Seeking Recommendation Model Based On Taxi GPS Data

Taxi is an important part of urban public transportation,but because most cruise taxi drivers were not clear about the distribution of passenger flow,they couldn't quickly and accurately find the hot spots of passenger demand,which led to the blind cruise phenomenon of many taxis and reduced the operational efficiency of taxi.Therefore,this paper identified the hot spots of passenger flow through the GPS data of taxi,established the taxi passenger seeking recommendation model,and provided guidance for taxi drivers to improve the efficiency of taxi operation.Firstly,this article built a Map Reduce processing framework to pre-process the GPS data of taxis.It used a geometric map matching algorithm to complete the map matching of GPS data.It designed a taxi passenger pick-up and drop-off location extraction algorithm and an OD matching algorithm.Secondly,a kernel density clustering algorithm was used to construct a density field passenger flow hotspot recognition model,and analyzed the spatial and temporal distribution of Ningbo residents' travel rules,taxi operating characteristics,and taxi passenger flow hotspot areas.The analysis showed that the peak hours for residents traveling by taxi on weekdays are 7:00-9:00,17:00-19:00 and 20:00-23:00.Residents travel more randomly on the rest days.The main service time zone for taxis was within 30 minutes,and the key service time zone was within 15 minutes.Transportation hubs,medical and health services,commercial services,and residential areas were hotspots for taxi demand.Among them,large-scale transportation hubs and medical and health hotspots have always maintained a high level;commercial service areas had higher hotspot ratings on rest days,while residential areas' hotspot rating was higher on weekdays.Finally,a taxi search recommendation model was established,which took the shortest travel time from the taxi search point to the hot spot area,the taxi supply and demand ratio in the hot spot area and the traffic performance index of road network as variables.The Dijkstra algorithm was used to calibrate the shortest travel time,calculated the taxi supply-demand ratio on the basis of regional division,and calculated the traffic performance index based on the travel time ratio.These methods were used to calibrate the model parameters.Applying this model to the guidance of taxi search for passengers in the morning rush hour on July 22,2017 in Ningbo,the analysis and comparison of the recommended results showed that the areas with the highest hotspot levels were not necessarily the areas with the highest recommendation for passenger search.It was closely related to the situation of taxi supply and demand in hotspot areas and the running status of road network traffic.The results showed that the model has a certain guiding role for taxi search.