Determining components of components of a PCA using princomp()

3 次查看（过去 30 天）
Tim
Tim 2012-6-3
Hi, I'm new to PCA so this might not be the right question. I'm trying to use princomp() to get a better understanding of what's driving my data set, and there are about 10 components that end up driving 99% of the variability. (It's a big data set, and I'm measuring growth over time periods specifically, hopefully to make the statistical analysis more rigorous)
However, I can't figure out what the output components translate to in my original data set, i.e. which combinations of variables they correspond to. Can someone help me figure out what these components are, or alternatively tell me what I'm getting wrong about PCA?
Thanks!

请先登录，再回答此问题。

回答（1 个）

the cyclist
the cyclist 2012-6-3
The description of the outputs is here: http://www.mathworks.com/help/toolbox/stats/princomp.html.
The first output, called "COEFF" in the documentation, is the matrix of coefficients of the principal components. Each column of COEFF identifies the linear combination of one of the principal components. So, for example, if the first column of COEFF is
>> [0.6435; 0.0102; 0.2754; ....]
then the first principal component is
P1 = 0.6435*x(:,1) + 0.0102*x(:,2) + 0.2754*x(:,3) + ...
etc.
Does that help?

类别

在 Help Center 和 File Exchange 中查找有关 Dimensionality Reduction and Feature Extraction 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