Multi-user model moving track prediction method based on sequential pattern mining

Abstract

The invention discloses a multi-user model moving track prediction method based on sequential pattern mining. Transverse track data positioned through mobile phone signaling is utilized to establish multiple user models to predict. According to the multi-user model moving track prediction method based on sequential pattern mining, a stay point mining algorithm with outlier tolerance is improved and provided, and stay points in the track are effectively mined; an inter-user similarity index based on a user stay area is provided, based on the index, transverse user track data is divided, and different user prediction models are established; based on a PrefixSapn algorithm, a moving track prediction method based on a sequential mining mode is provided, and the prediction model can be updated through newly generated real-time tracks.

Description

technical field [0001] The invention belongs to the technical field of data mining, and relates to a method for preprocessing mobile phone signaling traces based on clustering and outlier analysis. Background technique [0002] The user's current location is a key user context attribute, which can be associated with many applications to provide better location-based services (LBS Location Based Service). convenience. In the field of transportation, the real-time location information of vehicles has also become the basis for the implementation of congestion analysis and road flow analysis. Obviously, if the user's current location can be effectively obtained by some means, and its future location can be effectively predicted, its potential value is immeasurable. [0003] There have been a considerable amount of research on current trajectory prediction. From the way of data utilization, it is mainly divided into two categories, one is the vertical prediction method, and th...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that the prediction is based on the longitudinal historical data of a single user, and it cannot be predicted well in the case of less historical data.

[0009] Because people's mobile behavior has certain rules, that is, people tend to spend most of their time traveling to and from a few places with high frequency, so most domestic and foreign researches are based on longitudinal data to make predictions, and most Relying on complete, static, high-sampling rate user trajectory data, while mobile phone signaling trajectory has the characteristics of low sampling rate, less personal historical data but rich horizontal data, and the existing research methods are difficult to be directly applied to mobile phone positioning scenarios

Images