Understanding Taxi Service Strategies
Understanding Taxi Service Strategies
Taxi service strategies, as the crowd intelligence ofmassive taxi drivers, are hidden in their historical time-stampedGPS traces. Mining GPS traces to understand the service strate-gies of skilled taxi drivers can benefit the drivers themselves, pas-sengers, and city planners in a number of ways. This paper intendsto uncover theefficientandinefficienttaxi service strategies basedon a large-scale GPS historical database of approximately 7600taxis over one year in a city in China. First, we separate the GPStraces of individual taxi drivers and link them with the revenuegenerated. Second, we investigate the taxi service strategies fromthree perspectives, namely, passenger-searching strategies, passen-ger-delivery strategies, and service-region preference. Finally, werepresent the taxi service strategies with a feature matrix andevaluate the correlation between service strategies and revenue,informing which strategies are efficient or inefficient. We predictthe revenue of taxi drivers based on their strategies and achieve aprediction residual as less as 2.35 RMB/h,1which demonstratesthat the extracted taxi service strategies with our proposed ap-proach well characterize the driving behavior and performanceof taxi drivers.https://srisivasakthitravels.com/
smart phones and GPS navigators, a large number ofdigital footprints characterizing people’s mobility behaviorshave become available. These digital footprints provide us witha unique opportunity to understand human behaviors in varioussituations and exploit the underlying intelligence [7], [12], [15],[20], [32].In many cities, taxis are equipped with GPS devices, whichperiodically report to a central server the real-time informationabout the vehicle, including taxi locations and whether the taxiis occupied by a passenger. The collected GPS traces implicitlyconvey the service behaviors of the taxi drivers, includingwhere they pick up the passengers and how they find and deliverthe next passengers. These service behaviors vary from onedriver to another, depending on one’s service strategy in a givensituation. For example, after dropping off passengers, somedrivers may prefer to wait for new passengers at some familiarplaces, whereas others may prefer to search for new passengersin a busy area.The strategies employed by taxi drivers have a direct influ-ence on the amount of time and distance the taxi is occupied/vacant, resulting in differences in revenue, fuel consumption,and carbon emission. Good service strategies not only lead tohigh operating revenues but also improve efficiency of the entiretaxi service system and bring better services to passengers. Inaddition, good service strategies can help reduce carbon emis-sion by either decreasing the vacant/occupied driving distanceratio or selecting energy-efficient routes. Thus, understandingthe service strategies of taxi drivers can benefit the driversthemselves, passengers, and city planners.It is known that a number of factors affect the taxi drivers’service strategies, including passenger demands along a huntingroute, potential travel distance of passengers, waiting time,traffic conditions, competition from other taxis, and cost offuel. Most of the existing work intends to improve the taxiservice performance by building models to explicitly take intoaccount some of these factors. For instance, based on historicaltaxi GPS records, some recent work focuses on extractingpassenger pickup hotspots (i.e., location where the number ofpickup events is greater than a certain threshold) and urbanhuman mobility patterns in order to aid vacant taxis in findingthe next passengers [4], [13], [17], [21], [31]. Consideringthe competition introduced by other vacant taxis, Yuanet al.[31] propose to estimate the passenger-searching possibilities ofroads, which are used to infer the optimal passenger-searchinglocations and routes for drivers. Although an increasing numberof practical factors have been incorporated into the existingmodels, there is no analytical model proposed to systematicallystudy which service strategies are efficient/inefficient for bothpassenger searching and passenger delivery. click here to view full web site
https://goo.gl/maps/1rNVoifRFQ9snSor6
Taxi service strategies, as the crowd intelligence ofmassive taxi drivers, are hidden in their historical time-stampedGPS traces. Mining GPS traces to understand the service strate-gies of skilled taxi drivers can benefit the drivers themselves, pas-sengers, and city planners in a number of ways. This paper intendsto uncover theefficientandinefficienttaxi service strategies basedon a large-scale GPS historical database of approximately 7600taxis over one year in a city in China. First, we separate the GPStraces of individual taxi drivers and link them with the revenuegenerated. Second, we investigate the taxi service strategies fromthree perspectives, namely, passenger-searching strategies, passen-ger-delivery strategies, and service-region preference. Finally, werepresent the taxi service strategies with a feature matrix andevaluate the correlation between service strategies and revenue,informing which strategies are efficient or inefficient. We predictthe revenue of taxi drivers based on their strategies and achieve aprediction residual as less as 2.35 RMB/h,1which demonstratesthat the extracted taxi service strategies with our proposed ap-proach well characterize the driving behavior and performanceof taxi drivers.https://srisivasakthitravels.com/
smart phones and GPS navigators, a large number ofdigital footprints characterizing people’s mobility behaviorshave become available. These digital footprints provide us witha unique opportunity to understand human behaviors in varioussituations and exploit the underlying intelligence [7], [12], [15],[20], [32].In many cities, taxis are equipped with GPS devices, whichperiodically report to a central server the real-time informationabout the vehicle, including taxi locations and whether the taxiis occupied by a passenger. The collected GPS traces implicitlyconvey the service behaviors of the taxi drivers, includingwhere they pick up the passengers and how they find and deliverthe next passengers. These service behaviors vary from onedriver to another, depending on one’s service strategy in a givensituation. For example, after dropping off passengers, somedrivers may prefer to wait for new passengers at some familiarplaces, whereas others may prefer to search for new passengersin a busy area.The strategies employed by taxi drivers have a direct influ-ence on the amount of time and distance the taxi is occupied/vacant, resulting in differences in revenue, fuel consumption,and carbon emission. Good service strategies not only lead tohigh operating revenues but also improve efficiency of the entiretaxi service system and bring better services to passengers. Inaddition, good service strategies can help reduce carbon emis-sion by either decreasing the vacant/occupied driving distanceratio or selecting energy-efficient routes. Thus, understandingthe service strategies of taxi drivers can benefit the driversthemselves, passengers, and city planners.It is known that a number of factors affect the taxi drivers’service strategies, including passenger demands along a huntingroute, potential travel distance of passengers, waiting time,traffic conditions, competition from other taxis, and cost offuel. Most of the existing work intends to improve the taxiservice performance by building models to explicitly take intoaccount some of these factors. For instance, based on historicaltaxi GPS records, some recent work focuses on extractingpassenger pickup hotspots (i.e., location where the number ofpickup events is greater than a certain threshold) and urbanhuman mobility patterns in order to aid vacant taxis in findingthe next passengers [4], [13], [17], [21], [31]. Consideringthe competition introduced by other vacant taxis, Yuanet al.[31] propose to estimate the passenger-searching possibilities ofroads, which are used to infer the optimal passenger-searchinglocations and routes for drivers. Although an increasing numberof practical factors have been incorporated into the existingmodels, there is no analytical model proposed to systematicallystudy which service strategies are efficient/inefficient for bothpassenger searching and passenger delivery. click here to view full web site
https://goo.gl/maps/1rNVoifRFQ9snSor6
Comments
Post a Comment