matlab normalfit,求教copulafit

建议采Patton_copula_toolbox,

可以直接得到*negative* LL.

而copulafit则需修改程序,

我改为copulafit1.m,得到的nll当然

与Patton_copula_toolbox相同.

楼主若需要我再上传.

两者比较如下:

%%%%%%%%%

load stockreturns

x = stocks(:,1);

y = stocks(:,2);

u = ksdensity(x,x,'function','cdf');

v = ksdensity(y,y,'function','cdf');

options = statset('copulafit');

lowerBnd = options.TolBnd;

%%%%% 'Clayton'

theta0 = 1;

lower = 0.0001;

[kappa2,LL2] = fmincon('claytonCL',theta0,[],[],[],[],lower,[],[],options,[u,v])

%  kappa2 =    1.3979, LL2 = -25.1516

[alphaHat,nll]=copulafit1('Clayton', [u v])

%  alphaHat =  1.3979, nll = -25.1516

%%%%% 'Frank'

[ kappa5 LL5] = fmincon('frankCL',theta0,[],[],[],[],lower,[],[],options,[u,v])

%  kappa5 =    6.2835  LL5 = -34.1131

[alphaHat,nll]=copulafit1('Frank', [u v])

%  alphaHat =  6.2835  nll = -34.1131

%%%%% 'Gumbel'

[ kappa6 LL6] = fmincon('gumbelCL',theta0,[],[],[],[],lower,[],[],options,[u,v])

%  kappa6 =    2.0826, LL6 =  -35.5635

[alphaHat,nll]=copulafit1('Gumbel', [u v])

%  alphaHat =  2.0826, nll =  -35.5635