[程式] RATs跑DCC-GARCH無法收斂
------------------------------------------------------------------------
TITLE請著名軟體類別
按ctrl+v可預覽 發文前請把灰色的註解刪除
可使用ctrl+y刪除一整行
[軟體程式類別]:
請填入軟體程式類別 例如SAS、SPSS、R、EVIEWS...等
RATs
[程式問題]:
資料處理、迴歸、敘述統計、logistic、probit...等
DCC-GARCH無法收斂
MAXIMIZE - Estimation by BFGS
NO CONVERGENCE IN 185 ITERATIONS
LAST CRITERION WAS 0.0000000
SUBITERATIONS LIMIT EXCEEDED. ESTIMATION POSSIBLY HAS STALLED OR MACHINE
ROUNDOFF IS MAKING FURTHER PROGRESS DIFFICULT.
TRY HIGHER SUBITERATIONS LIMIT, TIGHTER CVCRIT, DIFFERENT SETTING FOR
EXACTLINE OR ALPHA ON NLPAR.
RESTARTING ESTIMATION FROM LAST ESTIMATES OR DIFFERENT INITIAL GUESSES
MIGHT ALSO WORK
[軟體熟悉度]:
低(1~3個月)
根據不同熟悉度
板上高手可以給予你不同指導
新手或許需要的還是語法的熟悉
而其他部分則可能往往是盲點上的點破
[問題敘述]:
請詳盡敘述遭遇到的問題，可能的話，分點敘述你要處理的流程
在進行DCC-GARCH的係數估計時,先以SIMPLEX估計起始值,
再以BFGS估計係數值,
但出現
MAXIMIZE - Estimation by BFGS
NO CONVERGENCE IN 185 ITERATIONS
LAST CRITERION WAS 0.0000000
SUBITERATIONS LIMIT EXCEEDED. ESTIMATION POSSIBLY HAS STALLED OR MACHINE
ROUNDOFF IS MAKING FURTHER PROGRESS DIFFICULT.
TRY HIGHER SUBITERATIONS LIMIT, TIGHTER CVCRIT, DIFFERENT SETTING FOR
EXACTLINE OR ALPHA ON NLPAR.
RESTARTING ESTIMATION FROM LAST ESTIMATES OR DIFFERENT INITIAL GUESSES
MIGHT ALSO WORK
請問該如何修正呢?
能夠分點敘述處理流程，基本上在撰寫程式或者跑統計時是相當有幫助的
甚至有經驗的人可以藉此告知你步驟是否錯誤
[程式範例]:
雖然張貼程式很可怕，但基本上有些程式還是要張貼才能解決
calendar(irregular)
allocate 2919
open data "d:\data-ln.txt"
data(format=free,org=cols) / lnftse lndax lnftsemib lnibex lncac lnstoxx lnsp
compute gstart=14,gend=2919
...
...
(1)MAXIMIZE(parmset=meanparms+garchparms,METHOD=simplex,ITERS=300,
subiters=100000)logl gstart gend
(2)MAXIMIZE(parmset=meanparms+garchparms,METHOD=bfgs,ITERS=300,subiters=100000,
cvcrit=0.000001) logl gstart gend
此時出現
MAXIMIZE - Estimation by BFGS
NO CONVERGENCE IN 185 ITERATIONS
LAST CRITERION WAS 0.0000000
SUBITERATIONS LIMIT EXCEEDED. ESTIMATION POSSIBLY HAS STALLED OR MACHINE
ROUNDOFF IS MAKING FURTHER PROGRESS DIFFICULT.
TRY HIGHER SUBITERATIONS LIMIT, TIGHTER CVCRIT, DIFFERENT SETTING FOR
EXACTLINE OR ALPHA ON NLPAR.
RESTARTING ESTIMATION FROM LAST ESTIMATES OR DIFFERENT INITIAL GUESSES
MIGHT ALSO WORK
於是我將(2)改成
MAXIMIZE(parmset=meanparms+garchparms,METHOD=bfgs,ITERS=300,subiters=100000000,
cvcrit=0.0000000001) logl gstart gend
結果出現
MAXIMIZE - Estimation by BFGS
NO CONVERGENCE IN 83 ITERATIONS
LAST CRITERION WAS 0.0000000
SUBITERATIONS LIMIT EXCEEDED. ESTIMATION POSSIBLY HAS STALLED OR MACHINE ROUNDOFF IS MAKING FURTHER PROGRESS DIFFICULT.
TRY HIGHER SUBITERATIONS LIMIT, TIGHTER CVCRIT, DIFFERENT SETTING FOR EXACTLINE OR ALPHA ON NLPAR.
RESTARTING ESTIMATION FROM LAST ESTIMATES OR DIFFERENT INITIAL GUESSES MIGHT ALSO WORK
是我修改方式錯誤嗎?
請問各位高手要如何修正呢?
非常謝謝!!!
-----------------------------------------------------------------------------
--
※ 發信站: 批踢踢實業坊(ptt.cc)
◆ From: 58.114.240.210
推
05/10 22:24, , 1F
05/10 22:24, 1F
→
05/10 22:24, , 2F
05/10 22:24, 2F
→
05/10 22:26, , 3F
05/10 22:26, 3F