Separate Gaussian mixture distributions of 5 components.

6 次查看（过去 30 天）
Amr Hashem
Amr Hashem 2017-4-8
评论： Amr Hashem 2017-4-10
I used this code (to fit data by gaussian mixture):
binWidth = 1; %width of the bar equal 1
binCtrs = 0:binWidth:134; % the bars start from 0 to 172. d=729 w=104 93 m=24
figure
counts = hist(Life,binCtrs,'k'); % b k m %hold on; plot(n,count,'ro')
stem(binCtrs,counts,'Marker','none','Color',[0.2 0.2 0.2]);
counts = hist(Life,binCtrs);
nu = length(Life);
prob = counts / (nu * binWidth);
figure
stem(binCtrs,prob,'Marker','none','Color',[0.5 0.5 0.5]);hold on ; % axis([-5 144 0 0.018]);
%%finding the number of components
%using minimum value of AIC
AIC = zeros(1,7);
obj = cell(1,7);
for Kk = 1:7
obj{Kk} = gmdistribution.fit(Life,Kk);
AIC(Kk)= obj{Kk}.AIC;
end
[minAIC,numComponents] = min(AIC);
numComponents %differ 5=4 7=6 11=5
%%finding the parameters of Mixture of gaussian using (gmdisribution.fit)
numComponents=4;
paramEsts= gmdistribution.fit(Life,numComponents)
%mu of mixture of gaussians of 4 components
MU=[paramEsts.mu(1);paramEsts.mu(2);paramEsts.mu(3);paramEsts.mu(4)];
SIGMA=cat(3,[paramEsts.Sigma(1)],[paramEsts.Sigma(2)],[paramEsts.Sigma(3)],[paramEsts.Sigma(4)]);%I wrote it at this
PPp=[paramEsts.PComponents(1),paramEsts.PComponents(2),paramEsts.PComponents(3),paramEsts.PComponents(4)];
objA = gmdistribution(MU,SIGMA,PPp);
xgridss=transpose(linspace(0,134,100)); %using the parameters to draw a line from 0 to 173
plot(xgridss,pdf(objA,xgridss),'g--','linewidth',2); hold on
I want to separate each gaussian component (I use this code):
numComponents5=4;
paramEsts5= gmdistribution.fit(Life,numComponents5)
MU1=[paramEsts5.mu(1)];
SIGMA1=cat(3,[paramEsts5.Sigma(1)]);
PPp1=[paramEsts5.PComponents(1)];
objA1 = gmdistribution(MU1,SIGMA1,PPp1);
xgridss1=transpose(linspace(0,134,100));
MU2=[paramEsts5.mu(2)];
SIGMA2=cat(3,[paramEsts5.Sigma(2)]);
PPp2=[paramEsts5.PComponents(2)];
objA2 = gmdistribution(MU2,SIGMA2,PPp2);
xgridss2=transpose(linspace(0,134,100));
MU3=[paramEsts5.mu(3)];
SIGMA3=cat(3,[paramEsts5.Sigma(3)]);
PPp3=[paramEsts5.PComponents(3)];
objA3 = gmdistribution(MU3,SIGMA3,PPp3);
xgridss3=transpose(linspace(0,134,100));
MU4=[paramEsts5.mu(4)];
SIGMA4=cat(3,[paramEsts5.Sigma(4)]);
PPp4=[paramEsts5.PComponents(4)];
objA4 = gmdistribution(MU4,SIGMA4,PPp4);
xgridss4=transpose(linspace(0,134,100));
% figure
plot(xgridss1,pdf(objA1,xgridss1),'r-','linewidth',2) ; hold on
plot(xgridss2,pdf(objA2,xgridss2),'r-','linewidth',2) ; hold on
plot(xgridss3,pdf(objA3,xgridss3),'r-','linewidth',2) ; hold on
plot(xgridss4,pdf(objA4,xgridss4),'r-','linewidth',2) ; hold on
but the value of each gaussian not the same How I could adjust the code to plot seprate individual components?

请先登录，再回答此问题。

回答（1 个）

the cyclist
the cyclist 2017-4-9
Do you mean you want each gaussian in a separate plot? Then,
figure
plot(xgridss1,pdf(objA1,xgridss1),'r-','linewidth',2);
figure
plot(xgridss2,pdf(objA2,xgridss2),'r-','linewidth',2);
figure
plot(xgridss3,pdf(objA3,xgridss3),'r-','linewidth',2);
figure
plot(xgridss4,pdf(objA4,xgridss4),'r-','linewidth',2);
Or perhaps use the subplot command.
If that is not what you mean, then it is not clear (to me) what you want.
  2 个评论
Amr Hashem
Amr Hashem 2017-4-9
编辑：Amr Hashem 2017-4-10
I am sorry if my question is not clear.
I have this (Gaussian mixture of 5 components)
and I want to get something like this:
which is five separate normal distributions in the same scale/Area (mu and sigma) fitting the data.
Thanks in advance.

请先登录，再进行评论。

类别

在 Help Center 和 File Exchange 中查找有关 Gaussian Mixture Distribution 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by