Effects of an outlying observation on multiple linear regression model : an empirical example in forward selection procedures.
When there are many candidates of predictor variables in a linear regression, often we have problem of selecting only important variables among them. Classical forward selection regression is often used with this goal but it could be very sensitive to the presence of an outlying observation. In this...
保存先:
| 主要な著者: | , |
|---|---|
| フォーマット: | Conference or Workshop Item |
| 言語: | English English |
| 出版事項: |
2012
|
| オンライン･アクセス: | http://psasir.upm.edu.my/id/eprint/27702/1/ID%2027702.pdf http://psasir.upm.edu.my/id/eprint/27702/ |
| タグ: |
タグ追加
タグなし, このレコードへの初めてのタグを付けませんか!
|
| 要約: | When there are many candidates of predictor variables in a linear regression, often we have problem of selecting only important variables among them. Classical forward selection regression is often used with this goal but it could be very sensitive to the presence of an outlying observation. In this paper, we show that the presence of a single outlier may irritate of the selecting variables in step(s) of the forward selection in linear regression. As a result, the final linear regression model will be wrongly specified which leads to have misunderstanding of decision makers. |
|---|