2014 Fiscal Year Final Research Report
Proximity data analysis for large and complex data
| Project/Area Number |
24500352
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Statistical science
|
|---|
| Research Institution | Doshisha University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HATANO Kenji 同志社大学, 文化情報学部, 准教授 (80314532)
FUKAGAWA Daiji 同志社大学, 文化情報学部, 助教 (10442518)
|
|---|
| Project Period (FY) |
2012-04-01 – 2015-03-31
|
| Keywords | ビッグデータ / シンボリックデータ / 多次元尺度構成法 / クラスタリング / 行列分解型多変量解析 |
|---|
| Outline of Final Research Achievements |
The proximity data is made of similarity or dissimilarity between two objects. The typical statistical methods to analyze the proximity data include Multidimensional Scaling (MDS ) and Clustering methods. However, since the data becomes larger and more complicated recently, sometimes the existing method does not provide the interpretable result and/or does not work because of the amount of computation.
Therefore, in this study, for large and complicated proximity data, we propose new statistical methods via an approach by symbolic data analysis, by using subspace, by simultaneous analysis with existing method and dimensional reduction method.
|
| Free Research Field |
多変量データ解析
|
