Assessing The Reliability Of Identifying Individual Stays From Mobile Phone Signaling Data

Mobile phone signaling data has become a widely used individual spatiotemporal positioning data.Recognizing individual staying patterns from mobile phone signaling data is the basis and key step of many researches on human spatiotemporal activities.Due to the problems of data noise and data sparseness in mobile phone signaling data,it is more difficult to identify individual stay patterns from mobile phone signaling data,and a series of research on mobile phone signaling data stay recognition algorithms have been produced.On the one hand,due to barriers such as data sampling restrictions and individual privacy restrictions,it is difficult to obtain verification data of the individual-level stay pattern corresponding to the mobile phone signaling data one-toone,resulting in a long-term reliability assessment of the individual stay recognition effect of the mobile phone signaling data.Difficulty.On the other hand,facing different subdivision application scenarios,it is necessary to consider the difference in the reliability evaluation of the stay recognition result.Therefore,this paper has carried out a research on the reliability assessment of individual stay recognition for mobile phone signaling data.The main contents include:(1)An individual-level stay recognition verification data set is constructed for mobile phone signaling dataThis research designed and implemented volunteer mobile phone signaling data and GPS travel data collection experiments,and constructed an individual-level mobile phone signaling data stay identification verification data set.This data set contains a total of 34 person-days of sample data.The analysis found that if the GPS data is used as the real location,the average distance deviation of the mobile phone signaling data is 1.4 kilometers.In addition,the distribution of the sampling frequency of mobile phone signaling data in the time dimension is heterogeneous.(2)Evaluated the reliability of the results of the representative stay recognition algorithm for the needs of differentiated applicationsThis study first screened the original data samples based on the two indicators of "spatial accuracy" and "temporal integrity".Then,in response to the application scenario requirements of mobile phone signaling data,this study proposed three types ranging from lenient to rigorous.The evaluation methods are "space discrimination","space-time state discrimination","space-time object discrimination",and corresponding evaluation indicators are set.Based on the data set and evaluation system constructed above,this study evaluated the results of three representative stay recognition algorithms SMo T,SMUo T,and TSC-MAD.The results show that SMUo T has the best overall performance under the three evaluation methods;the spatial recognition threshold of the stay recognition algorithm has a more significant impact on the stay results than the temporal recognition threshold;When the consideration of residence time is added to the evaluation method,the evaluation performance of the three algorithms is significantly worse,recognition of individual residence time is the difficulty of individual stay recognition.To sum up,this research explored the construction method of the individual stay recognition verification dataset of mobile phone signalingdata,and provided an individual stay recognition evaluation method,which is widely based on mobile phone signaling data.The research and application of signaling data stay recognition provides a basic reference.