Evaluation and Reconstitution of Existing Models of Urban Structure in Japan Using Machine Learning

2023 Fiscal Year Final Research Report

• Project/Area Number: 19K21671
• Research Category: Grant-in-Aid for Challenging Research (Exploratory)
• Allocation Type: Multi-year Fund
• Review Section: Medium-sized Section 4:Geography, cultural anthropology, folklore, and related fields
• Research Institution: Kyoto Sangyo University (2023); Kogakkan University (2019-2022)
• Principal Investigator: KIRIMURA Takashi 京都産業大学, 文化学部, 准教授 (70584077)
• Project Period (FY): 2019-06-28 – 2024-03-31
• Keywords: 都市圏 / 都市内部構造モデル / 小地域統計 / 地理情報システム（GIS） / 機械学習

Outline of Final Research Achievements

This study aims to quantitatively evaluate the suitability of the traditional urban internal structure models for Japanese cities using machine learning techniques, as well as to construct a new urban internal structure model through the classification of urban internal structure using machine learning for Japanese cities of various sizes. Although the quantitative evaluation using machine learning did not necessarily produce satisfactory results, certain results were achieved in terms of the development of a regional classification method using supervised classification and a new urban internal structure analysis method using a machine learning approach as an alternative to the factorial ecological approach.

Free Research Field

都市地理学

Academic Significance and Societal Importance of the Research Achievements

本研究は、近年急速に発達してきた機械学習手法を、まだ十分な適用事例の蓄積がない都市地理学的課題に適用しようとしたものである。当初の研究計画は十分には達成できなかったものの、リモートセンシングで用いられている教師付き地域分類を都市内部構造の分析に利用し、類型や因子へのラベル付け問題への対処を図ったほか、従来用いられてきた因子分析主体の居住地域構造分析に非負値テンソル因子分解（NTF）などの機械学習手法を用いたことも本研究の重要な達成点である。