作者查詢 / PinkQQ

總覽項目：發文 | 留言 | 暱稱

作者 PinkQQ 在 PTT [ Statistics ] 看板的留言(推文), 共9則

限定看板：Statistics

看板排序：

首頁

尾頁

[ Statistics ]9 留言, 推噓總分: +1

作者: candypinky - 發表於 2009/05/14 12:15(15年前)

1^F推PinkQQ: simply put, one could apply GLS while the problem of07/07 13:55

2^F→PinkQQ: heteoskedastic errors occur, so that we want to find07/07 13:55

3^F→PinkQQ: a nonsingular and squared matrix to transform the data07/07 13:56

4^F→PinkQQ: However, GLS is often unable to applied because we do07/07 13:56

5^F→PinkQQ: not know what Var(y) is (so does the transformed M).07/07 13:57

6^F→PinkQQ: Hence, we use FGLS instead. But another problem is07/07 13:57

7^F→PinkQQ: that FGLS often doesn't have the nice properties in07/07 13:58

8^F→PinkQQ: finite sample and could be a less efficient estimator07/07 13:58

9^F→PinkQQ: than OLS.07/07 13:58

首頁

尾頁