Programación genética

La regresión simbólica

Abstract

Regression analysis is a statistical analysis that aims to deduct the pattern in a series of data or research the statistical relation between a dependent variable (Y) and one or more dependent variables, the result is an algebraic expression type Y=F (X1, X2, …Xn). This article has the most common regression analysis: lineal regression which has one independent variable Y=F(X). A common user comes into contact with lineal regression when using electronic sheets that implement tendency line deduction given a series of data. However, he/she will notice there are certain limits to this technique for example, the data has sinusoidal behavior or follows some algebraic function behavior or a combination of algebraic functions beyond the offered menu: lineal, polynomial, potential, logarithmic or exponential. Symbolic regression (a genetic programming application) has the same objective as lineal regression but with a much greater search spectrum and much less limitations: Given the data, it will search for the pattern (algebraic expression) that identifies their behavior ascending to all types of functions and algebraic combinations.