MATLAB: Which components are kept in PCA with Classification Learner?

After getting the optimal classification model in Clasification Learner, I added a PCA to see if it would improve the accuracy. Thus, it gives the following result:

• PCA is keeping enough components to explain 90% variance.
• After training, 18 components were kept.
• Explained variance per component (in order): 22.5%, 10.3%, 8.4%, 6.6%, 5.7%, 4.8%, 4.4%, 3.9%, 3.5%, 3.2% (variances of least important components hidden).

I would really need to know WHICH are the 18 components it selects as well as to which components each of the percentages of explained variance belong to.

Thank you very much to all of you!

Answer 1

[Drew]

Each PCA component consists of a linear combination of the original features.

The PCA components are ordered according to the explained variance, so it selects the top 18 PCA components. Each of those 18 PCA components is calculated as a linear combination of the original features, as seen in the PCA matrix.

See https://www.mathworks.com/help/stats/pca.html for more information

Answer 2

[Image Analyst]

@Ariadna Colmenero Cobo de Guzmán, how many variables did you start with? 18 I imagine. Were there more?