Symbolic regression is an algorithm that finds an equation best fitting given data by searching through a space of possible mathematical expressions.
Unlike classical regression methods that use a predefined model structure, symbolic regression freely searches for formulas combining addition, multiplication, powers, and other operations.
Typically, this search is conducted using evolutionary algorithms or genetic programming, where formulas are represented as trees and optimized through variation.
This method enables the discovery of interpretable models and is especially useful when the underlying relationships are unknown or complex.