How to apply pca() [Matlab] on high dimensional data

14 visualizaciones (últimos 30 días)
Atinesh Singh el 14 de Ag. de 2016
Comentada: Akash Reddy el 10 de Nov. de 2020
I want to apply `pca()` function in `matlab` on data with `500 dimensions`. But pca() has a limit of only 99 dimensions. Do I have to write code for pca.
0 comentariosMostrar -2 comentarios más antiguosOcultar -2 comentarios más antiguos

Iniciar sesión para comentar.

the cyclist el 14 de Ag. de 2016
Why do you believe pca has such a limit?
p = pca(rand(1000,700));
runs just fine.
4 comentariosMostrar 2 comentarios más antiguosOcultar 2 comentarios más antiguos
the cyclist el 17 de Ag. de 2016
I think I finally appreciate what you are missing.
You have more dimensions (p=700) than you have observations (n=100). When p>n, you can fully explain all the variation in the observations with n-1 principal components, which in your case is 99.
Akash Reddy el 10 de Nov. de 2020
Thanks sir.

Iniciar sesión para comentar.

Más respuestas (2)

John D'Errico el 14 de Ag. de 2016
Editada: John D'Errico el 14 de Ag. de 2016
I think the problem is you don't understand the PCA code, at least how to use the tool as provided. READ THE HELP! A problem with size 100x700 for the PCA function is a problem with 700 dimensions, not 100. PCA treats each ROW of the array as one sample, one observation.
Your question (coupled with your later comment) strongly implies that your array is simply transposed from what you need to pass into the PCA tool. Read the help for PCA.
2 comentariosMostrar NingunoOcultar Ninguno
Atinesh Singh el 16 de Ag. de 2016
I've read the documentation. It's clearly mentioned that pca() takes input a matrix of dimension n-by-p where n = # observations and p = # variables and return a matrix with dimension p-by-p where each column is a principal component with decreasing variance. Hence when we pca() on matrix with dimension 100-by-700 it just return matrix with dimesnion 700-by-700.
Walter Roberson el 16 de Ag. de 2016
It is not clear to me why you think that pca has a limit of 99 dimensions?
I had no problem at all a moment ago running pca on a 1000 x 1000 matrix.

Iniciar sesión para comentar.

Taimour Hamayoun el 25 de Oct. de 2017
hello! i am new in matlab and i am try to apply PCA on my dataset of 19 dimensions and try to reduce it in 4 dimension but i didnt find the proper way plz guide and provide me a proper source with explanation thanx
1 comentarioMostrar -1 comentarios más antiguosOcultar -1 comentarios más antiguos
the cyclist el 27 de Oct. de 2017
May I suggest that you carefully read the documentation and this answer of mine, to get a better understanding of the syntax and output of pca?
Also, you posted this as an answer to a question. It would have garnered more attention as a new question, but I happened to see it.

Iniciar sesión para comentar.

Categorías

Más información sobre Dimensionality Reduction and Feature Extraction en Help Center y File Exchange.

Community Treasure Hunt

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

Start Hunting!

Translated by