総合的類型化解析基盤の構築とその応用

摘要

現在，行政，医療，学校など様々な分野で既存要素の統廃合が進んでいる.また今後予想されるさらなる高齢化社会，若年者人口減少，労働者の不足などを大きな社会問題として抱えている.そのような問題に対応するために，各分野での類型化を実施して対象毎の特徴を明らかにし，効率的な改善が必要である.またビッグデータ環境が広がり，多くのデータを取り込み，解析を実施するためには，整備された解析環境が必要である，またオープンデータやWebAPIを利用することで，新しい要素を取り入れた解析が可能になる.本研究ではこのようなビッグデータ環境の中で類型化を実施するための類型化解析基盤の構造およびフローを整理し，任意の規模で類型化を実施できる環境を提案する.類型化にはDEAを用い，各要素の特徴を明確にするとともに改善指標を示し，結果を可視化する.その類型化解析基盤を用いて，実際にある医療機関の類型化を実施し，特徴を持った分類と現実的な時間内での計算を確認する.%There is an ongoing effort to consolidate existing elements in various fields such as government, medicine, and education. At the same time, society is facing considerable challenges, such as a projected aging of the population, a decline in the younger population, and a shortage of workers. To respond to these issues, it is necessary to classify the features of the objects in each field and to propose efficient improvements. In this research, we propose a structure and flow for a classification analysis foundation, as well as an environment in which classifications can be performed on any scale. A data envelopment analysis (DEA) is used for the classification in order to clarify the features of each element. We then use the analysis foundation to classify medical and educational institutions in order to investigate the type of categories produced. Various approaches have been used for classification. Methods used in cluster analyses include the farthest neighbor method and the k-means method. However, because these methods use average values, it is possible that the features of individual elements may remain unclear. Moreover, classification foundations include a geographically specialized foundation, which does not lend itself to general use. Furthermore, many classification analyses restrict the number of objects, casting doubt on their ability to cope with large-scale classification analyses. In this research, we propose an analysis foundation that is capable of general-purpose, large-scale classifications. The classification analysis foundation is structured to enable the classification of many objects, including medical institutions, educational institutions, and patient analyses. Here, APIs are used to provide the subject data for the classifications, and a DEA is used for the calculations to clarify the features of the objects.