Taximeter Fraud Detection Based On GPS Trace Data

Taxies play important role in transportation especially in today’s city daily life.In recent decades,the invention of electronic taximeter make the taxi trip go with much more simple and fair charging way.Unfortunately,though taximeters has provide citizens with great convenience,some greedy drivers may modified the taximeter to acquire more profits,which is unfair to passengers.For government,the most effective existing supervision method is to randomly select some taxies to and check the rotate speed of the taximeter manually.It is obviously with low cost in time and manpower.Since GPS trace has been utilized in various area such as city dynamic,we can also leverage it to detect taximeter fraud.In fraud taxi,drivers modify reported task distance,which will lead us a larger average speed as baseline.Good news is the instant speed reported by GPS device is reliable.We propose a system named VOut,which suggests our solution are based on the abnormal record may have more speed be out of the normal range.We evaluate the scheme in the case of diverse parameters and get some well-performance empirical value combination in each case.The metric of accuracy is false positive rate and false negative rate.We implement one candidate method based on map-matching and quartile to compare with VOut.