- [Home]
- [Research achievement]
- [Research achievement detail]
| Title | Bayes Prediction Algorithm for Regression Model with Outliers (in Japanese) |
|---|---|
| Authors | Tota Suko 、Fumitaka Nakagawa 、Toshiyasu Matsushima |
| Released Year | 2004 |
| Format | Conference |
| Category | Others |
| Jounal Name | Proceedings The Seventh Workshop on Information-Based Induction Sciences |
| Jounal Page | pp.34-39 |
| Published Year | 2004 |
| Published Month | 11 |
| Abstract （English） |
In statistics analysis, the outliers of the data affect the analysis result.Therefore, there are many research for analysis in the case of the data including outliers.The analyses are categorized two types, the methods which assume models which include outliers, or not include outliers.And, It is central subject that outlier detection in analyze statistical data which include outliers.But, there is no theoretical guarantee that prediction methods detecting and getting rid of outliers.Then, in this study, we propose an optimal prediction method with reference to the Bayes criterion in regression models which consider occurence of outlier. The computational complexity of this method grows exponentially.And so, we propose an approximation method for reducing the computational complexity, and estimate this method through some simulations. |
| Note （English） |
1 |
| Manuscript | |
| Presentation |
