Principal components vs Make principal components
Posted: Tue Jul 20, 2021 5:04 am
Hello,
I have a question related to PCA. I am trying to make an index based on the first principal component. Apparently, Eviews has a nice built-in function for that in "Proc>Make principal components" but I have a couple of questions before I can proceed with the output analysis.
1) I wonder if I need to normalize my data before it or the option "Normalize loadings" does it automatically? I am trying to avoid double normalization.
2) Another question is how to find a linear representation of the new index by the factors? I tried to reestimate it manually by taking PC1 loadings of each factor from the "View>Principal components" but the two do not match.
I have a question related to PCA. I am trying to make an index based on the first principal component. Apparently, Eviews has a nice built-in function for that in "Proc>Make principal components" but I have a couple of questions before I can proceed with the output analysis.
1) I wonder if I need to normalize my data before it or the option "Normalize loadings" does it automatically? I am trying to avoid double normalization.
2) Another question is how to find a linear representation of the new index by the factors? I tried to reestimate it manually by taking PC1 loadings of each factor from the "View>Principal components" but the two do not match.