調査と因子分析によるエージェントベースモデリング

―消費者のデジタルコンテンツ選択行動を事例に―

抄録

 If the agent base model is classified based on the granularity, it can be classified into three types. A simple abstracted simple model is an “Abstract Model.” “Facsimile Model” maps a limited reality society and details are described. “Middle Range Model” is constructed with the abstraction degree between the above two types. The importance of “Facsimile Model” has been pointed out due to sophis- tication of information technology, and improvement of methodology to describe agent attributes and behavioral guidelines are desired to construct a more realistic model.

 In this research, we perform a questionnaire survey to acquire single source data, and an estima- tion of digital content consumer parameters is attempted using factor analysis that is one of multivar- iate analysis.

 Conclusions of this study are as follows. (1) Perform a questionnaire survey to gather single source data and analyze the data by factor analysis are helpful to construct an agent-based model. (2) After the agent-based modeling, a hybrid-gaming simulation is proper to verify the reasonability of the model (3) To fit the consumer parameters to sales of digital contents, another analysis to classify the attribute of contents is needed.