Detecting a local hot spot using a varying coefficient model for spatial-temporal data and its application to survival data

2020 Fiscal Year Final Research Report

• Project/Area Number: 17K00052
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Statistical science
• Research Institution: Shiga University (2019-2020); Hiroshima University (2017-2018)
• Principal Investigator: Satoh Kenichi 滋賀大学, データサイエンス教育研究センター, 教授 (30284219)
• Co-Investigator(Kenkyū-buntansha): 冨田 哲治 県立広島大学, 地域創生学部, 教授 (60346533)
• Project Period (FY): 2017-04-01 – 2021-03-31
• Keywords: 成長曲線モデル

Outline of Final Research Achievements

In Satoh and Tonda (Applied Statistics, 2013), the proposed method of Brumback et al. (JASA, 1999) was applied to longitudinal measurement data. This research was highly evaluated and received the 2015 Society of Applied Statistics Award (Best Paper Award). Tonda, Satoh et al. (Applied Statistics, 2010) proposed an application to spatial data. And Satoh & Tonda (JJSS, 2014) proposed a semiparametric inference of change coefficient surfaces for spatial data. And the main results of this research were summarized in Satoh (Applied Statistics, 2020), "Application of Balanced Growth Curve Model to Spatio-Temporal Data.

Free Research Field

統計学

Academic Significance and Societal Importance of the Research Achievements

位置情報を持つ個体において経時測定データが観測されている場合に，時間と空間の交互作用項を持つ成長曲線モデルを適用することを試みる．位置や時間の基底について考え，それぞれを固定した場合の時間軸上の予測曲線あるいは空間上の予測曲面に関する同時信頼区間を与える．また，交互作用項，特に空間に関する基底数を減らすための工夫として局所的な成長曲線モデルの適用も考える．提案するモデルの適合は重相関係数の意味でとても良かった．