Symbolic regression by uniform random global search

• Published in: SN applied sciences Vol. 2; no. 1; p. 34
• Main Author: Towfighi, Sohrab
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01.01.2020; Springer Nature B.V
• Subjects: 3. Engineering (general); Applied and Technical Physics; Chemistry/Food Science; Datasets; Earth Sciences; Engineering; Environment; Experiments; Generalized linear models; Genetic algorithms; Materials Science; Mathematical analysis; Optimization algorithms; Regression; Research Article; Software; Source code; Symbolic regression; Tree enumeration; Evolutionary algorithm; Genetic programming; 62-07; Uniform random global search; Pure random search; 05C05
• ISSN: 2523-3963; 2523-3971; 2523-3971
• DOI: 10.1007/s42452-019-1734-3

This work describes a novel algorithm for symbolic regression, namely symbolic regression by uniform random global search (SRURGS). SRURGS has only one tuning parameter and is very simple conceptually. The method produces random equations, which is useful for the generation of symbolic regression benchmark problems. We have released well documented and open-source python code which passed a formal peer-review, so that interested researchers can deploy the tool in practice. We conduct experiments comparing SRURGS with symbolic regression by genetic programming (SRGP) using 100 randomly generated equations. Our results suggest that SRGP is faster than SRURGS in producing equations with good R 2 for simple problems. However, our experiments suggest that SRURGS is more robust than SRGP, able to produce good output in more challenging problems.