Standardization transforms every feature to mean 0 and standard deviation 1. Equalizing feature spread prevents large-scale features from dominating principal component analysis (PCA), while centering simplifies its computation. section 6.2.3
Within a PCA workflow, standardization precedes the following operations:
- Compute the covariance matrix from the standardized features. section 6.2.3
- Find principal components as covariance-matrix eigenvectors, ranked by their eigenvalues. section 6.2.3
- Retain the first k components and represent each sample in the resulting lower-dimensional coordinate system. section 6.2.3
Python PCA workflow
A representative implementation creates a StandardScaler before PCA, then configures PCA to retain enough components to explain 99% of the variance. section 6.2.3
Figure 6.9 PCA defines a new coordinate system based on principal components, where the important dimensions are self-evident, and we can reduce the data dimensionality to k by retaining the first k PCs as ranked by the amount of variance they explain.

View how PCA creates a new coordinate system and selects components by explained variance. section 6.2.3






