The most common linear methods for feature extraction are Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) . PCA uses an orthogonal transformation to convert data into a lower-dimensional space while maximizing the variance of the data.