The "Ping-pong Effect" is a typical noise data in mobile phone location data. It is of great significance to analyze its typical spatial and temporal characteristics to scientifically deal with the influence of noise data, accurately understand the analysis results and improve the quality of mobile phone location data. With mobile phone location data from three different cities (Shanghai, Xining and Shenzhen), three detection methods were used to identify the "Ping-pong Effect" in mobile phone data, and to compare and analyse its jump distance characteristics, spatial distribution characteristics and temporal distribution characteristics. The results show that: In terms of spatial distribution, Ping-pong Effects shows a high spatial autocorrelation phenomenon, showing that it is easy to occur in distant suburban areas and not easy to occur at high base station density; in terms of temporal distribution characteristics, it shows that it is easy to occur during daytime and not easy to occur at night; the existence of the "Ping-pong Effect" will have a certain degree of influence on the research results.
关键词
手机位置数据/"乒乓效应"/时空数据挖掘/数据质量
Key words
mobile phone location data/"Ping-pong Effect"/spatial-temporal data mining/data quality
万方数据