| Massive mobile phone location data(MPLD),which contains richfull human activity information,have been widely used to understand human activity patterns.However,this type of data source is not collected for human mobility studies.The temporal resolution varies across the datasets in existing studies.The varying temporal resolutions results in two kinds of issues that affect the outcomes of the human mobility studies derived from mobile phone location data:(1)the decline of the temporal resolutions leads to the decrease of the effectiveness traditional method(e.g.,speed based method)to detect the Ping-phong phenomenon,which is a typical noise data in mobile phone location data;(2)the varying temporal resolutions lead to the temporal aggregation effect,which indicates that the outcomes of spatio-temporal analysis relies on the temporal resolution.This dissertation aims to develop new methods to reduce the impacts cuased by the undetect Ping-pong phenomenon and investigate how the temporal aggregation effect affect the outcomes of human mobility studiesFirst,Ping-pong phenomenon is a typical noise in mobile phone location data.It exhibits quick and frequent jumps between several nearby locations.Speed-based method is commonly-used in detecting Ping-pong phenomenon in previous studies.However,its effectiveness will decrease significantly with the decline of temporal resolutions(e.g.,bigger than five minutes).This study proposed new methods to detect Ping-pong phenomenon based on its abnormal movement speed,repeat jumps between fixed locations,frequent turns,as well as the combination method based on the above three characteristics.The proprosed method exhibits higher effectiveness in detecting Ping-pong phenomenon when the temoral resolutions is relative coarse(e.g.,15 minutes).This study further investigates the characteristics of the spatial and temporal dstributions and jump distances of Ping-pong phenomenon using a mobile phone location data of Shanghai.The results indicate that(1)the Ping-pong phenomenon is easiers to happen during 19:00～23:00 and(2)less to happen in the downtown area;(3)the jump distanes of more than 40%Ping-pong phenomena are further than 500 meters.Moreover,this study also finds that the Ping-pong phenomena in call detail records(CDRs)datasets have minor impacts in evaluating the mobility patterns from the population perspective but have complex impacts in evaluating the mobility pattrens for each individual.Second,identifying stops is a primary step in acquiring activity-related information from mobile phone location data to understand the activity patterns of individuals.However,Ping-pong phenomenon in mobile phone location data may create "fake moves," which will generate fake activity patterns of "stops-and-moves."These "fake moves" share similar spatiotemporal features with real short-distance moves,and the stops and moves of trajectories(SMoT),which is the most extensively used stop identification model,often fails to distinguish them when the dataset has coarse temporal resolutions.This study proposes the stops,moves,and uncertainties of trajectories(SMUoT)model to address this issue by introducing uncertain segment analysis to distinguish "fake moves" and real short-distance moves.A real mobile phone location dataset collected in Shenzhen,China is used to evaluate the performance of SMUoT.The results indicate that SMUoT improves the performance(i.e.,28 and 22%increase in accuracy and recall rate for a half-one-hour temporal resolution dataset,respectively)of stop identification and exhibits high robustness to parameter settings.Third,the temporal resolutions of MPLD determines how well such data can describe human activities and influence the values of human mobility indicators.However,systematic investigations of how the temporal resolutions affects human mobility indicators remain lacking.To fill this gap,this study uses a MPLD containing 19,370 intensively sampled individual trajectories(temporal resolution is samller than 5 minutes)to systematically assess the impacts of the temporal resolutions on four selected mobility indicators that describe human mobility patterns from different aspects,specifically movement entropy,the radius of gyration,eccentricity,and daily travel frequency.The results indicate that:(1)the radius of gyration,which mainly relies on the spatial extent of the locations of activities with long duration times,is stable with increasing temporal resolutions.Whereas,movement entropy exhibits a slight higher underestimation than that of eccentricity,while obvious lower underestimation than that of daily travel frequency.(2)The underestimation of movement entropy among individuals show high consistency,while those of eccentricity and daily travel frequency exhibit significant inconsistency.The change patterns of these indicators are related to the way of their calculations as well as the activity locations and duration times of individuals.The above findings can help to better understand and interpret the results derived from MPLD with a certain coarse TSI.It also reminds researchers to choose proper temporal resolutions to answer particular questions.Last,the impacts on human mobility studies caused by temporal aggregation effect varies across populations with different activity patterns.The main reason is that the spatial and temporal information of an activity with long duration is relatively easier to be caught in a trajectory with a fixed temporal resolution,while a short-lasting activity is relatively easier to be neglected.Therefore,this study further compares the impacts of temporal aggregation effect on human mobility indicators between populations with different activity patterns.(1)A mathmethic model is built to estimate the type and the probability of an activity with a certain duration time in a trajectory with a given temporal resolutions.This model can help design appropriate temporal resolutions when collecting MPLD or evaluate the potential biases in a given MPLD with a certain temporal resolution in activity-oriented applications.(2)This study evaluates the impacts caused by temporal aggregation effect on three mobility indicators(including movement entropy,radius of gyration and eccentricity)between poplulations with different activity patterns devided based on daily travel distance(DTD)and daily travel frequency(DTF).The results indicate that:(i)people in HH group(DTD>9 km and DTF>4 times)receive two times higher underestimations for their movement entropy than that of people in LL group(DTD<9 km and DTF<4 and DTF>0).In contrast,their radius of gyration and eccentricity receives only 1/3 underestimation than those of people in LL group.(ii)People in the special group whose activity space extent is smaller than 500 meters exhibit similar patterns of the impacts from TSIs with LL group.However,their eccentricities receive significant higher underestimations those of all other groups.(iii)For radius of gyration and movement entropy,the underestimations of the whole populations are close to those of the HH group,but further from those for LL group and the special group.Findings in this study help researchers to learn how TSIs affect the human mobility indicators for populations with different activity patterns.Potential social equality concerns may arise if these datasets are misused. |
PDF Full Text Request