# How do I perform unbalanced ANOVA ?

135 views (last 30 days)
Simo on 22 Oct 2014
Answered: Hanling Su on 10 Feb 2020
Hi! I have 3 vectors of data (each vector has a different number of data) and I'd like to perform ANOVA. How do I do it?

the cyclist on 23 Oct 2014
If you have the Statistics Toolbox, you can use the anovan function to do unbalanced ANOVA.

Samar Ragab on 22 Sep 2019
try this ,
g1=ones(1,size(classA3,1));
g2=2*ones(1,size(classB3,1));
g3=3*ones(1,size(classC3,1));
p1=zeros(1,26);
for i=1:26 %%%crossponding to features numbers(columns)
x=[classA3(:,i);classB3(:,i);classC3(:,i)]' ;
g=[g1,g2,g3];
p1(i)=anova1(x,g,'off');
end

Hanling Su on 10 Feb 2020
Hey you can use anova1 but assign specific group variable, for example, you can create two vector, the first one is the observations of your sample, group by group data = [x11, x12, ... x120, x21, x22, ... x227, x31, x32, .. , x317]. And you create a correspongding group vector, for this example group = [1,1,1...,1, 2, 2, ... , 3, 3, ...], as long as the observations have the same index as its group, you can use anova1(data, group), it will automatically group the data for you.

Translated by