[問題] 關於stata,multinomial logit,找PredictP
又來請教板上的大家了QQ
想問當我利用stata進行mlogit後
想要得到Predict Probability的問題
第一個是outcome A的整理
. tab A
A | Freq. Percent Cum.
------------+-----------------------------------
1 | 501 62.78 62.78
2 | 245 30.70 93.48
3 | 52 6.52 100.00
------------+-----------------------------------
Total | 798 100.00
而當我跑完molgit之後
依照stata手冊上所說想得到估計的機率
. predict P1 P2 P3
(option pr assumed; predicted probabilities)
. summarize P1 P2 P3
Variable | Obs Mean Std. Dev. Min Max
-------------+--------------------------------------------------------
P1 | 798 .6278195 .0961358 .2042496 .8797781
P2 | 798 .3070176 .1041065 .0910776 .7751256
P3 | 798 .0651629 .017927 .0193713 .1271836
得到的各個比例卻跟實際樣本的比例(tab中)一樣
62.78 vs 0.6278195
30.70 vs 0.3070176
6.52 vs 0.0651629
想問說這個predict是代表說把樣本再放到跑完mlogit之後的估計式中得到的嗎?
而當我計算margin effects時也會得到一個predict possibility感覺比較像要求的
. mfx, predict (outcome(1))
Marginal effects after mlogit
y = Pr(A==1) (predict, outcome(1))
= .63693578
------------------------------------------------------------------------------
variable | dy/dx Std. Err. z P>|z| [ 95% C.I. ] X
---------+--------------------------------------------------------------------
CE | .005055 .00159 3.18 0.001 .001939 .008171 49.4875
CG | .0121339 .00584 2.08 0.038 .000685 .023583 12.896
CJ | .0004675 .00027 1.73 0.083 -.000062 .000997 95.5702
------------------------------------------------------------------------------
表裡面的 y = Pr(A==1) (predict, outcome(1)) 感覺比較像我得到的
= .63693578
不知道用margin effect所做的是否正確
然後如果想得到估計的機率要怎麼求得呢
想確定一下兩種方式估計出來的機率是代表甚麼意思
再次感謝大家QQ
--
※ 發信站: 批踢踢實業坊(ptt.cc)
◆ From: 140.112.86.49
※ 編輯: myjstyle 來自: 140.112.86.49 (03/14 14:13)
推
03/14 14:31, , 1F
03/14 14:31, 1F
→
03/14 14:33, , 2F
03/14 14:33, 2F
→
03/14 14:33, , 3F
03/14 14:33, 3F
→
03/14 14:34, , 4F
03/14 14:34, 4F
→
03/14 14:34, , 5F
03/14 14:34, 5F
※ 編輯: myjstyle 來自: 140.112.86.49 (03/14 14:34)
推
03/14 14:38, , 6F
03/14 14:38, 6F
→
03/14 14:39, , 7F
03/14 14:39, 7F
→
03/14 14:39, , 8F
03/14 14:39, 8F
→
03/14 14:39, , 9F
03/14 14:39, 9F
→
03/14 14:40, , 10F
03/14 14:40, 10F
→
03/14 14:41, , 11F
03/14 14:41, 11F
→
03/14 14:41, , 12F
03/14 14:41, 12F
→
03/14 14:42, , 13F
03/14 14:42, 13F
推
03/14 14:42, , 14F
03/14 14:42, 14F
→
03/14 14:42, , 15F
03/14 14:42, 15F
→
03/14 14:43, , 16F
03/14 14:43, 16F
→
03/14 14:44, , 17F
03/14 14:44, 17F
→
03/14 14:44, , 18F
03/14 14:44, 18F
→
03/14 14:49, , 19F
03/14 14:49, 19F
→
03/14 14:49, , 20F
03/14 14:49, 20F
推
03/14 14:50, , 21F
03/14 14:50, 21F
→
03/14 14:51, , 22F
03/14 14:51, 22F
→
03/14 14:51, , 23F
03/14 14:51, 23F
→
03/14 14:51, , 24F
03/14 14:51, 24F
→
03/14 14:51, , 25F
03/14 14:51, 25F
→
03/14 14:52, , 26F
03/14 14:52, 26F
→
03/14 14:53, , 27F
03/14 14:53, 27F
推
03/14 14:54, , 28F
03/14 14:54, 28F
→
03/14 15:00, , 29F
03/14 15:00, 29F