Abstract
Knowing the behavioral patterns of city residents is of great value in formulating and adjusting urban planning strategies, such as urban road planning, urban commercial development, and urban pedestrian flow control. Based on the high penetration rate of cell phones, it is possible to indirectly understand the behavior of city residents based on the call records of users. However, the behavioral patterns of large‐scale users over a long period of time can present characteristics such as large dispersion, difficult to discover patterns, and difficult to explain behavioral patterns. In this paper, we design and implement a human behavior pattern analysis system based on massive mobile communication data based on serial data modeling method and visual analysis technology. For the problem that it is difficult to capture the behavioral patterns of residents in cities in call records, this paper constructs base station trajectories based on users’ cell phone call records and uses users’ long‐time base station trajectories to mine users’ potential behavioral patterns. Since users with similar activity characteristics will exhibit similar base station trajectories, this paper focuses on the similarity between text sequences and base station trajectory sequences and combines the word embedding method in natural language processing to build a Cell2vec model to identify the semantics of base stations in cities. In order to obtain the group behavior patterns of users from the base station trajectories of group users, a user clustering method based on users’ regional mobile preferences is proposed, and the results are projected using the Stochastic Neighbor Embedding (t‐SNE) algorithm to expose the clustering features of large‐scale cell phone users in the low‐dimensional space. To address the problem that user behavior patterns are difficult to interpret, a visual analysis model with group as well as regional semantics is designed for the spatial and temporal characteristics of user behavior. Among them, the clustering model uses the distance between scatter points to map the similarity between users, which helps analysts to explore the behavioral characteristics of group users.
Highlights
With advances in science and technology and computer hardware, today’s computer systems are capable of storing massive amounts of data
In order to present the similarity between users’ behavioral patterns and facilitate the selection of user groups of interest, this paper uses the t-distributed neighborhood embedding algorithm (t-SNE) algorithm to cluster users based on their regional movement preference vectors. t-SNE algorithm’s projection results can effectively present the similarity between data points, and this method is mainly used for dimensionality reduction and exploration of highdimensional data, which is an effective method for visualizing and presenting high-dimensional data
Users move in the city for a long time showing significant regularity, and when the number of user groups is large enough, their behavioral patterns can well reflect the population mobility of city residents
Summary
With advances in science and technology and computer hardware, today’s computer systems are capable of storing massive amounts of data. Researchers at Berkeley University estimate that the world generates about one million terabytes of data per year, mostly stored in digital form. As urbanization accelerates worldwide, people around the world are constantly generating, disseminating, and exchanging massive amounts of highdimensional heterogeneous data, from human social data involving microblogs, WeChat, and QQ to communication data generated by cell phone calls. Human behavior data has become one of the most common types of data used in urban visual analytics. The study of crowd flow patterns facilitates the development and adjustment of business marketing strategies, such as the accurate placement of advertisements and the precise location of businesses, promoting business profitability [2]. For urban management, understanding the gathering and movement patterns of crowds can help urban planners to formulate urban planning strategies and make timely road risk assessments and urban emergency predictions to adjust urban traffic strategies and urban emergency plans [3]
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call