Study On Demand Of Taxi Based On Didi And POI Data

As information era coming,the study on Big Data have been one of Research hotspots in academic and industry filed.Big Data have some advantages,such as variety data forms,high-volume,convenient and quickly data transmission.Big Data have already influenced and changed the thinking way of problems solutions.Besides data primitive value,people are also concerned about the deep knowledge beneath the covering raw data meanwhile,even more than.The association rules of different data forms,data classification,data clustering,outlier detection are important methods in data mining.The paper is the taxi demand study combine POI and didi taxi demands.Cities have variety POIs,such as shops,supermarkets,companies,scenic spots,residential districts,schools and so on.These POIs reflect the inner urban functions.Didi taxi demand reveals the taxi demand quantity which people want to take a taxi in a city.The taxi demand's changes in Different time periods give to administrator helpful suggestion of taxi distribution in proper district,which satisfy people taxi demand.Making use of taxi demand quantity,the paper studied on taxi demand aggregated district,taxi demand active time periods,the association feature and distribution about POIs in each district,the reasons of taxi demand changes in time periods and geographical locations on account of the POI features around taxi demand locations.1)The author distributed the time periods about taxi demand data.In the first,the author divided the time into weekday and weekend in a week,then divided the time in a day into four time periods,morning peak between 7am and 9am,afternoon time between 12 and 13 o'clock,evening peak between 17 and 18 o'clock,night time between 21 and 23 o'clock.Next,a taxi demand is equaled as the max value in 500 meters circle,and the taxi hour is same as the max taxi demand point's hour.The author got the demand distribution features by use of Kernel Density Estimation.When the demand were more than some value, the author would get the district which taxi demand quantity is larger.The district showed taxi demand's active time periods in time,active geographic location in geography.2)In active taxi demand area,the author analyzed whether taxi demand point have five kinds POI in 100 meters.The five points included supermarkets,companies,scenic spots,residential districts and schools.Next,making use of Apriori method,they would mine the association rules between taxi demand and POIs.Then,the statistics about POI numbers around taxi demand point showed realtions between taxi demand quantity and time periods,geographic location,weekend or weekday.And the author discussed the change reasons of time periods,geographic location,and active time periods in weekday and weekend.