Construction of nonparametric statistical theory on Network

2016 Fiscal Year Final Research Report

• Project/Area Number: 24500339
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Statistical science
• Research Institution: Kanazawa University
• Principal Investigator: SAGAE Masahiko 金沢大学, 経済学経営学系, 教授 (20215669)
• Research Collaborator: SCOTT Warren, David Rice University Department of Statistics
• Project Period (FY): 2012-04-01 – 2016-03-31
• Keywords: ノンパラメトリック統計学 / 方向統計学 / ネットワーク統計学

Outline of Final Research Achievements

We construct a statistical theory that can deal with data defined on the network.Network data has branches, closed circuits, etc. Therefore, we can not simply apply statistical theory on Euclidean space. We applied the directional statistical theory applicable to both closed and branched cases to this problem. Since parametric statistical theory is extremely limited in handling problems, we introduced nonparametric statistical theory to directional statistics. We led to a theoretical properties on directional density estimation that previous researches in this area were theoretically undeveloped. Specifically, we derive the asymptotic property of the wrapped Cauchy kernel. We proposed a new class of kernel directional density function and derived its asymptotic properties. The convergence rate of the bandwidth estimation method (CV method and plug - in method) of the direction kernel function was derived.

Free Research Field

ノンパラメトリック統計学