Error message while Using PCA function in Matlab
5 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
Skirt Zhang
el 19 de Ag. de 2014
Respondida: Skirt Zhang
el 9 de Ag. de 2017
Dear All,
I am using the matlab existing function> PCA to compute the PCs for matrix X (26, 5000), so 26 observations and 5000 variables.
But matlab gives the warning message as below @@Warning: Columns of X are linearly dependent to within machine precision. Using only the first 243 components to compute TSQUARED.
Does this mean the function only takes the first 243 variables to compute the PCs?
1 comentario
Biraj Bhandari
el 29 de Jul. de 2017
https://stackoverflow.com/questions/27997736/matlab-warns-columns-of-x-are-linearly-dependent-to-within-machine-precision
Respuesta aceptada
John D'Errico
el 29 de Jul. de 2017
Editada: John D'Errico
el 29 de Jul. de 2017
NO. The warning does NOT mean what you asked. There is no problem.
With only 26 data points, there are only at MOST 26 independent columns anyway. That it went down to 243 components only means that the author was conservative in the coding. A component is a linear combination of your variables. So they could have arguably stopped at the first 26 linear combinations and have been safe in the analysis.
So don't worry. The PCA will be entirely valid, using all 5000 variables, finding linear combinations of those variables.
0 comentarios
Más respuestas (1)
Ver también
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!