作者anovachen (囧)
看板Statistics
标题[程式] SAS proc autoreg 错误讯息 (TGARCH)
时间Thu Jun 20 20:35:35 2013
[软体程式类别]:
SAS 9.3
[程式问题]:
以autoreg进行TGARCH(1,1)模型建立。
[软体熟悉度]:
高(1年以上)
[问题叙述]:
由於EGARCH和TGARCH都无法收敛，
所以加入maxiter=1000强迫他多迭代几次，
总算估计出系数了...
但是，TGARCH部分仍有警告讯息：
ERROR: QUANEW Optimization cannot be completed.
NOTE: Optimization routine cannot improve the function value.
WARNING: Optimization cannot improve the function value.
请问这对於系数的估计会产生什麽影响？
[程式范例]:
proc autoreg data=data outest=garch_family maxiter=1000;
garch_1_1 : model r = / garch=(p=1,q=1); output out=out_garch
predicted=predicted residual=residual ht=ht lcl=lowerbound ucl=upperbound;
igarch_1_1 : model r = / garch=(p=1,q=1,type=integ); output
out=out_igarch predicted=predicted residual=residual ht=ht lcl=lowerbound
ucl=upperbound;
egarch_1_1 : model r = / garch=(p=1,q=1,type=egarch); output
out=out_egarch predicted=predicted residual=residual ht=ht lcl=lowerbound
ucl=upperbound;
garchm_1_1 : model r = / garch=(p=1,q=1,mean=log); output
out=out_garchm predicted=predicted residual=residual ht=ht lcl=lowerbound
ucl=upperbound;
qgarch_1_1 : model r = / garch=(p=1,q=1,type=qgarch); output
out=out_qgarch predicted=predicted residual=residual ht=ht lcl=lowerbound
ucl=upperbound;
tgarch_1_1 : model r = / garch=(p=1,q=1,type=tgarch); output
out=out_tgarch predicted=predicted residual=residual ht=ht lcl=lowerbound
ucl=upperbound;
pgarch_1_1 : model r = / garch=(p=1,q=1,type=pgarch); output
out=out_pgarch predicted=predicted residual=residual ht=ht lcl=lowerbound
ucl=upperbound;
run;
-----------------------------------------------------------------------------
後来改用trust region法估计系数，
即加入optmethod=tr指令，可以估出TGARCH的系数，
但是...改用这种数值最佳化法OK吗?
--
※ 发信站: 批踢踢实业坊(ptt.cc)
◆ From: 111.255.10.225
※ 编辑: anovachen 来自: 111.255.10.225 (06/21 00:46)