I have a large data set with many features, I want to know which ones are more important. I know there are some features selection method, but I am wondering does pca say some thing about it? for example if the coefficient of first component of a variable is larger then it has meaning?
PCA doesn't give you information about individual features per se, but you can plot the projection of the features onto PCA space, much like Anh Lee does in his pokemon data analysis, to get an idea of which features contribute most to the variance.
EDIT: You might also find this interesting.