Call Records Data Visual Analytics Towards Users Behavior Understanding

With the city's rapid developing,the highly frequently usage of mobile users made the services of mobile phones are becoming diversity and widely.Mobile users will generate a large amount of usage data during use,such as calls,Internet access and others.To analyzing the data of mobile phone usage,the analyst can not only describe a user but also can help operators adjust business strategies and provide auxiliary information for the construction and adjustment of communication facilities.Due to the mobile phone user's call record contains both time and space information about human activities and social relationship information between users,it has important value for human behavior analysis and understanding.Using traditional machine learning and pattern recognition methods made difficult and timeconsuming to detect the relationship between user behavior patterns and users,and the results lack explanatory evidence,and it is difficult for analysts to reasonably interpret the results.In view of the above problems,this paper attempts to use the user call records to extract user behavior characteristics to identify user relationships,and optimize the operation facilities based on the user's long-term behavior characteristics,mainly focusing on the following three aspects:1.Users characteristics and relations between others are implicated in large-scale user call records but the features about complex dimension relations and big scale made it hard to handle.This paper designs a method to explore and understand user behavior patterns through social and spatiotemporal information.This method can extract time and space,social information and other information from user call data.On the one hand,it recognizes the user's social role according to the long-term behavior characteristics of the household,and On the one hand,the community discovery algorithm is used to discover communities in the social network,and visual analysis methods are used to identify user relationships and potential interaction patterns between users.2.Use the visual analysis methods to identify user relationships and explore potential interaction patterns between users alone,will against the large-scale objects and lowly efficiency of task completion.To improve it,this work proposed a semi-automatic method to identify user relationships and potential interaction patterns.This method treats the user as a node,and the various types of interaction between users are regarded as the edges between the nodes.A multi-layer dynamic network is constructed,and the call and meeting relationship between users can be used as a sub-layer in the multi-layer network.The tensor decomposition method is used to identify the potential interaction patterns,and finally,the visualization model is designed to explain the results.3.Due to the development and vicissitude happened in urban all day,the functional areas and call needs in the city will also change.The mobile operator wants to obtain the current operating status and regional requirements of the base station network,and adjust the base station network accordingly,thereby solving the situation that the base station facilities cannot meet the usage requirements and the base station resources are wasted.In order to help this paper propose a visual analysis tool based on perceptual cycle theory and context + focus technology,it is used to explore and analyze the base station network from the perspective of time and space,which helps the analyst to focus on the area of interest without losing the context information.The analyst can complete those analysis tasks(location selection for cell station and network fine-tuning)in the interaction process.