Developer Guide for Intel® Data Analytics Acceleration Library 2019 Update 4
The metrics are computed given the input data meets the following requirements:
The PCA algorithm receives input argument eigenvalues
. It represents the following quality metrics:
The library uses the following quality metrics:
Quality Metric |
Definition |
---|---|
Explained variance |
![]() |
Explained variance ratios |
![]() |
Noise variance |
![]() p r - number of principal components, p - number of features in the data set |