Statistical network data analysis: Use of Tashu data

Abstract

Basically network data consists of nodes and edges. Each node represents each entity and edge represents relationship between nodes. More speci- cally, depending on how to express the relationship between the connected nodes, the binary edges take only 0 or 1, and the weighted edges take non- negative integers as their values. This paper deals with a series of processes from nding characteristics of weighted network data to selecting a model that best describes it. Based on the theoretical background, we will analyze the actual data of Tashu, which is unattended public bicycle rental system. In model selection, Valued ERGM will be used because the Tashu data has weighted network structure. Finally, after interpreting the results of analyses, limits and future topics of the research is discussed as a conclusion.