In the answers to the PCA open question on the 2021 exam normalization is said to refer to making the mean of each feature 0. In the lecture nodes it is stated that "Experience shows that is a good idea to remove the mean of each feature and to normalize the variance to one." Why do we not normalize the variance in the exam solution and only set the mean to zero (which I would probably refer to as centering rather than normalization)?
Normalization in PCA [exam 2021]
In the answers to the PCA open question on the 2021 exam normalization is said to refer to making the mean of each feature 0. In the lecture nodes it is stated that "Experience shows that is a good idea to remove the mean of each feature and to normalize the variance to one." Why do we not normalize the variance in the exam solution and only set the mean to zero (which I would probably refer to as centering rather than normalization)?
4
I think they just gave an example with 0 :)
When you normalize with the variance XX.T it gives the correlation matrix
Add comment