Introduction

[Research Content] With the advancement of data collection technologies, a wide variety of data has become available, but statistical science is an indispensable tool for deriving valuable insights from data. My research focuses on the problems encountered in actual data analysis and the development of new methodologies to solve them, as well as verifying their theoretical validity. In particular, I am interested in developing methodologies using the framework known as Bayesian statistics and statistical methodologies for analyzing data accompanied by time and location information. I have also conducted applied research in fields such as economics, medicine, public health, and environmental science. For specific research achievements, please refer to the link to my personal page.
[For Students Wishing to Join the Seminar] In my seminar, we broadly cover the techniques of statistical science and data science, with a focus on Bayesian statistics. You are free to pursue themes of interest, not only Bayesian statistics but also machine learning, spatiotemporal statistics, causal inference, etc. In the seminar, we emphasize not only understanding the theoretical aspects of analytical methods but also implementing and applying these methods concretely using R and Python. We welcome participants who want to deeply study statistical science and data science, research new analytical methods and theories, or conduct applied research using various data.

■Bayesian Methods
・Research Objective: To propose a Bayesian quantile trend filtering method for estimating the non-stationary trend of quantiles on graphs and apply it to crime data in Tokyo.
- Title: Locally adaptive spatial quantile smoothing: Application to monitoring crime density in Tokyo, Onizuka Takahiro, Hashimoto Shintaro, Sugasawa Shonosuke （Spatial Statistics） 59 100793 Mar. 2024
・Research Objective: To propose a new Bayesian method with global-local shrinkage priors for estimating monotone function values that is adaptive to local abrupt changes.
- Title: Locally adaptive Bayesian isotonic regression using half shrinkage priors, Okano Ryo, Hamura Yasuyuki, Irie Kaoru, Sugasawa Shonosuke （Scandinavian Journal of Statistics） 51 （ 1 ） 109-141 Mar. 2024
・Research Objective: To propose a Bayesian quantile trend filtering method using calibrated variational approximations for estimating non-stationary trends of quantiles.
- Title: Fast and locally adaptive Bayesian quantile smoothing using calibrated variational approximations, Onizuka Takahiro, Hashimoto Shintaro, Sugasawa Shonosuke （Statistics and Computing） 34 （ 1 ） Feb. 2024
■Nonparametric and Semiparametric Methods
・Research Objective: To develop a functional version of trend filtering for estimating the trend of functional data indexed by time or on a general graph.
- Title: Trend filtering for functional data, Wakayama Tomoya, Sugasawa Shonosuke （Stat） 12 （ 1 ） Jan. 2023
・Research Objective: To propose semiparametric Bayesian estimation for nonignorable missing data response mechanisms using penalized spline and radial basis function methods.
- Title: Bayesian semiparametric modeling of response mechanism for nonignorable missing data, Sugasawa Shonosuke, Morikawa Kosuke, Takahata Keisuke （TEST） 31 （ 1 ） 101-117 Mar. 2022
■Statistical Inference and Estimation
・Research Objective: To develop robust statistical inference methods for meta-analysis that adjust the influences of outlying studies using generalized likelihoods based on the density power divergence.
- Title: Robust inference methods for meta‐analysis involving influential outlying studies, Noma Hisashi, Sugasawa Shonosuke, Furukawa Toshi A. （Statistics in Medicine） 43 （ 20 ） 3778-3791 Sep. 2024
・Research Objective: To develop improved methods for constructing more accurate prediction intervals in network meta-analysis.
- Title: Improved methods to construct prediction intervals for network meta‐analysis, Noma Hisashi, Hamura Yasuyuki, Sugasawa Shonosuke, Furukawa Toshi A. （Research Synthesis Methods） 14 （ 6 ） 794-806 Nov. 2023
■Spatial and Longitudinal Data Analysis
・Research Objective: To propose a method for aggregating global and local sub-models for scalable and flexible spatial regression modeling using eigenvector spatial filtering.
- Title: Sub‐Model Aggregation for Scalable Eigenvector Spatial Filtering: Application to Spatially Varying Coefficient Modeling, Murakami Daisuke, Sugasawa Shonosuke, Seya Hajime, Griffith Daniel A. （Geographical Analysis） 56 （ 4 ） 768-798 Oct. 2024
■Robustness and Sensitivity Analysis
・Research Objective: Development of a fully data-driven normalized and exponentiated kernel density estimator using the Hyvärinen score.
- Title: Fully Data-Driven Normalized and Exponentiated Kernel Density Estimator with Hyvärinen Score, Imai Shunsuke, Koriyama Takuya, Yonekura Shouto, Sugasawa Shonosuke, Nishiyama Yoshihiko （Journal of Business & Economic Statistics） 43 （ 1 ） 110-121 Jan. 2025
・Research Objective: To obtain new sufficient conditions for posterior robustness in Bayesian linear regression using a contamination model with a light-tailed and heavy-tailed component.
- Title: Posterior robustness with milder conditions: Contamination models revisited, Hamura Yasuyuki, Irie Kaoru, Sugasawa Shonosuke （Statistics & Probability Letters） 210 110130 Jul. 2024

Areas of Research

・Bayesian Statistics
・Spatip-temporal Data Analysis
・Statistical Inference
・Robust Statistics

Social Contributions

・Scientific Advancement: New statistical methods enhance data analysis accuracy across various scientific fields.
・Data Analysis Enhancement: Improved statistical techniques lead to better decision-making in fields like finance and healthcare.