Symbolic Regression

Symbolic regression is a form of regression analysis that searches the space of mathematical expressions themselves, rather than fitting parameters to a single predetermined formula, to find the model that best balances accuracy and simplicity against a dataset. Because it infers both the structure and the parameters of the model directly from the data, it explores a far larger space of possible models than conventional regression, at the cost of the guidance that domain knowledge would otherwise provide. The most common technique for searching this space is genetic programming, which recombines candidate expressions, though newer approaches include Bayesian methods and neural network based methods such as AI Feynman. Its fitness function weighs accuracy against complexity to avoid overfitting, making symbolic regression valuable both for building predictive models and for producing human-readable equations that reveal the underlying structure of data. This description is adapted from Wikipedia contributors under CC BY-SA 4.0; changes were made. https://creativecommons.org/licenses/by-sa/4.0/