A form of compression that applies a mathematical transformation to the original data to reduce redundancy.