506 / 2019-02-28 23:31:42

Research on Spatial Position Accuracy of Mo-bile Phone Signaling Data

Mobile phone signaling data; position accuracy; spatial location; location inference

全体主题 > Big data in Urban Planning and Management

In recent years, mobile phone signaling data has increasingly become an important data type for the research on urban-rural space and resident behavior. Due to the lacking of investigations on the positioning accura-cy of mobile phone signaling data, this study collects both GPS location data and mobile signaling data of Shanghai tourists. Using the GPS data as the benchmark for accuracy verification, the study compares a linear regression model and an artificial neural network model. The tourists’ ac-tual locations are inferred from the cell stations using both models. Based on the four spatial scales of the neighborhood, traffic analysis zone, sub-district and administrative district, the study investigates the accuracy of the model estimations under each spatial scale. It finds that: (1) 86% of the cell stations are not the one closest to the actual locations; (2) the ar-tificial neural network model is a better for estimating the actual posi-tions than the linear regression model, and its accuracy reaches 70% at least. It is concluded that the neural network model could be used as a re-liable method for correcting the mobile phone signaling positioning data.