Mining And Analyzing Of User Mobile Patterns Based On Mobile Location Data

With the development of network technology like mobile network,Internet and Internet of Things, more and more spatiotemporal data produced by human, for example, GPS data of taxi, checking data from network media and mobile phone's location data, are perceived and collected. Because the spatiotemporal data contains trace information that can describe one's move, we can mine features and regular patterns of users.Further, interactive patterns can be obtained, which increasingly attracts attention from researchers. The field of mobile phone's location data,which is easily-available, widely-spread and informative, have been a study hotspot. Besides, operators can improve the ability of market response and client service innovation by exploring these data's additional value, finally, increasing their operating incomes and market competitive advantage.This thesis will focus on event-driven data (calls, messages) and network-driven data in telecommunications network, analyzing features and patterns of user behaviors. The main innovation and work are mainly reflected in the following aspects:1. Distributed processing framework based on Spark. This thesis will design a distributed processing framework based on Spark. The framework's basic functions include exception processing of mobile location data, cell oscillation resolution, stay point extraction and path generation.2. The discovery of user's important location. Based on the result of stay point extraction and path generation, this thesis will take statistic and unsupervised clustering method to identify user's residence and work place and greedy algorithm to find user's usual route. Furthermore, DBSCAN algorithm is accomplished based on Spark, which is used to identify important places in people's lives.3. User's mobile pattern extraction. With the help of Motif model in graphic mining, we mine user's mobile pattern from mobile phone's location data, and compare the difference between Hainan natives and foreign visitors.4. Behavior analysis of foreign visitors in Hainan. We adapt analytic hierarchy process and extract the mobile features of foreign visitors in Hainan. Further, we successfully classify visitors into common travel and business travel.