Updated on 2026/02/02
博士(工学) ( 2003.3 東京大学 )
Natural Science / Applied mathematics and statistics
Informatics / Statistical science
Research Organization of Information and Systems The Institute of Statistical Mathematics
Niigata University Faculty of Education, Chair of Mathematical and Natural Sciences
Niigata University Institute of Humanities and Social Sciences, Academic Assembly Professor
2025.4
Niigata University Fundamental Sciences, Graduate School of Science and Technology Professor
2025.4
Niigata University Mathematical and Natural Sciences, Faculty of Education Professor
2025.4
Niigata University Faculty of Education Mathematical and Natural Sciences Associate Professor
2014.4 - 2025.3
Niigata University Graduate School of Education School Subjects Associate Professor
2014.4 - 2025.3
On distance functions in multiply robust estimation of population means
Tadayoshi Fushiki
Statistical Papers 67 ( 1 ) 2026.1
A multiple imputation method using population information
Tadayoshi Fushiki
Communications in Statistics - Theory and Methods 54 ( 12 ) 3476 - 3493 2025.6
A note on convergence of calibration weights to inverse probability weights
Tadayoshi Fushiki
Statistica Neerlandica 79 ( 1 ) 2025.2
A note on the properties of estimators in missing data analysis
Tadayoshi Fushiki
Communications in Statistics - Theory and Methods 51 ( 17 ) 6144 - 6149 2022.9
On the Selection of the Regularization Parameter in Stacking
Tadayoshi Fushiki
Neural Processing Letters 53 ( 1 ) 37 - 48 2021.2
Nonresponse Bias Adjustment in Regression Analysis
Tadayoshi Fushiki, Tadahiko Maeda
Journal of Statistical Theory and Practice 14 ( 2 ) 2020.6
Tadayoshi Fushiki, Tadahiko Maeda
Behaviormetrika 41 ( 1 ) 99 - 114 2014.1
Estimation of prediction error by using K-fold cross-validation
Tadayoshi Fushiki
Statistics and Computing 21 ( 2 ) 137 - 146 2011.4
Bayesian bootstrap prediction
Tadayoshi Fushiki
Journal of Statistical Planning and Inference 140 ( 1 ) 65 - 74 2010.1
Estimation of positive semidefinite correlation matrices by using convex quadratic semidefinite programming
Tadayoshi Fushiki
Neural Computation 21 ( 7 ) 2028 - 2048 2009.7
A maximum likelihood approach to density estimation with semidefinite programming
Tadayoshi Fushiki, Shingo Horiuchi, Takashi Tsuchiya
Neural Computation 18 ( 11 ) 2777 - 2812 2006.11
Bootstrap prediction and Bayesian prediction under misspecified models
Tadayoshi Fushiki
Bernoulli 11 ( 4 ) 747 - 758 2005.8
Nonparametric bootstrap prediction
Tadayoshi Fushiki, Fumiyasu Komaki, Kazuyuki Aihara
Bernoulli 11 ( 2 ) 293 - 307 2005.4
On parametric bootstrapping and Bayesian prediction
Tadayoshi Fushiki, Fumiyasu Komaki, Kazuyuki Aihara
Scandinavian Journal of Statistics 31 ( 3 ) 403 - 416 2004.9
A phenomenon like stochastic resonance in the process of spike-timing dependent synaptic plasticity
Tadayoshi Fushiki, Kazuyuki Aihara
IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences E85-A ( 10 ) 2377 - 2380 2002.10
A Study of the Japanese National Character: Succession and Development
Grant number：23H00062
2023.4 - 2027.3
System name：Grants-in-Aid for Scientific Research
Research category：Grant-in-Aid for Scientific Research (A)
Awarding organization：Japan Society for the Promotion of Science
Grant amount：\48750000 （ Direct Cost: \37500000 、 Indirect Cost：\11250000 ）
A study on variable selection in nonresponse adjustment
Grant number：15K00043
2015.4 - 2019.3
System name：Grants-in-Aid for Scientific Research
Research category：Grant-in-Aid for Scientific Research (C)
Awarding organization：Japan Society for the Promotion of Science
Fushiki Tadayoshi
Grant amount：\2340000 （ Direct Cost: \1800000 、 Indirect Cost：\540000 ）
In order to investigate the properties of variable selection methods for nonresponse adjustment techniques, real data analysis and computer simulation were conducted. Several variable selection methods were compared by real data. The results showed that they did not affect the estimates as much as expected. Computer simulation studies showed that unnecessary auxiliary variables do not affect the estimates too much in a special situation.
A Study on the quality control of interviewer-mediated surveys using survey paradata and methods of nonresponse bias adjustment
Grant number：15H03424
2015.4 - 2018.3
System name：Grants-in-Aid for Scientific Research
Research category：Grant-in-Aid for Scientific Research (B)
Awarding organization：Japan Society for the Promotion of Science
Maeda Tadahiko, KIKKAWA Toru, KATO Naoko
Grant amount：\16120000 （ Direct Cost: \12400000 、 Indirect Cost：\3720000 ）
This study examined the survey paradata, which mainly means data about the survey process obtained in the administration of surveys, for the purpose of improving the quality of survey operation by using this information. We also discussed the methods for evaluating nonreseponse bias which could be caused by low response rates, and methods for adjusting for the bias. Survey modes included in this study were traditional face-to-face interviewing with paper questionnaire, self-administered questionnaire, telephone interview by RDD, Web surveys and on-site smart-card record of visitor's behavior. By analyzing the visit record in interviewer-mediated surveys and call record of RDD surveys, we can understand the interviewer behavior more precisely and we can make use of these findings in interviewer training. By analyzing paradata such as response time in CAPI survey or Web survey, we can deepen our understandings on the respondent behavior.
A study on Bayesian prediction and bootstrap prediction
Grant number：20700260
2008 - 2009
System name：Grants-in-Aid for Scientific Research
Research category：Grant-in-Aid for Young Scientists (B)
Awarding organization：Japan Society for the Promotion of Science
FUSHIKI Tadayoshi
Grant amount：\1300000 （ Direct Cost: \1000000 、 Indirect Cost：\300000 ）
The problem to predict future observations based on past observations is one of the problems that are widely interested in statistics. In this study, we clarified the relation between bootstrap prediction and Bayesian prediction and calculated the predictive performances of them, both theoretically and experimentally. For real data analysis, we developed a method for evaluating the predictive performance.
アンサンブル法の統計的予測問題への適用
Grant number：17700286
2005 - 2007
System name：科学研究費助成事業
Research category：若手研究(B)
Awarding organization：日本学術振興会
伏木 忠義
Grant amount：\1800000 （ Direct Cost: \1800000 ）
これまでと同様に,Kullback-Leiblerダイバージェンスを損失関数とした統計的予測問題を考えた.昨年,サンプル数がモデルの大きさに比べて大きいとはいえない状況で,ブートストラップ予測を構成するときに問題が生じることを示し,その問題を解決する方法を考えた.具体的には,Rubinが提案したベイジアン・ブートストラップを用いて予測分布を構成する手法を提案した.昨年度は,ベイジアン・ブートストラップを用いた予測分布について,漸近理論を用いて理論解析を行うとともに,簡単なモデルを使って理論の確認を行ったが,本年度は実データを用いて現実的な状況でその有効性を調べた.Boston郊外の家の値段を,その地域の犯罪率,ある広さ当たりの住居地の占める割り合い,街に占める小売店以外の会社の広さの割り合いといった量をもとにして予測するBoston Housing Dataなどのデータを用いて,ベイジアン・ブートストラップ予測,ブートストラップ予測,プラグイン予測の予測性能の比較を行った.複雑な現象を扱う場合には,大きなモデルを使う必要があるが,サンプル数とパラメータ数が近い状況となる.そのような状況ではブートストラップ予測では問題が生じることがあり,ベイジアン・ブートストラップ予測の安定性が確認された.漸近理論を用いたブートストラップ予測のプラグイン予測に対する予測の改良分は2次のオーダーであり,データ数が大きな場合は小さな量となると考えらけるが,このような状況では予測の改良分は大きく,本手法の有効性が確認された.また,本年度は,これらの結果をまとめ,論文として投稿した.
数学・数学教育学研究入門
微分積分学I
統計学特講
ベイズ統計学概論
応用解析学II
統計学特論Ⅰ
応用解析学I
スタディ・スキルズH
小学校算数
数学科教材開発研究特論
解析学特論II
統計学II
卒業研究
統計学I
情報数学II
情報数学I
くらしと数理
情報教育論
