← back

principal component analysis

click to drop points — toggle to project into PC space

How it works

Center the data by subtracting the mean. Compute the covariance matrix. Find its eigenvectors — those are the principal components.

PC1 points in the direction of maximum variance. PC2 is orthogonal and captures the rest. Arrow length scales with standard deviation.

Toggle to pca space to see the data projected onto these axes — each point's (PC1, PC2) score.

PC1 variance
PC2 variance