标题: A Genetic Programming-Driven Data Fitting Method [打印本页] 作者: 杨利霞 时间: 2020-11-12 16:33 标题: A Genetic Programming-Driven Data Fitting Method
A Genetic Programming-Driven
. t5 {- H' q2 b& J, t
Data Fitting Method
f# ^# w( u$ I/ \) p% a- v) y
0 R" B4 e2 w1 Y- h! @* y$ d/ l) A: q! [6 [# E' c! p
0 N& w0 b+ c' H- u- o: ]* pData fifitting is the process of constructing a curve, or a set of mathematical functions, that has0 G* G) y! w) O
the best fifit to a series of data points. Different with constructing a fifitting model from same type of function,8 a. T) }+ n# |
such as the polynomial model, we notice that a hybrid fifitting model with multiple types of function may have 6 }" w/ C. D2 V+ X' _) W0 y% Qa better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid ; j) d9 v; d5 }' ?8 I; Wfifitting model depends on a reasonable combination of multiple functions and a set of effective parameters. - a. l, f. J% z' N8 {That is a high-dimensional multi-objective optimization problem. This paper proposes a novel data fifitting 5 C. @' _& b: [9 M7 w! vmodel construction approach. In this approach, the model is expressed by an improved tree coding expression F! i- R! N8 u s* E' Aand constructed through an evolution search process driven by the genetic programming. In order to verify ! x M# k! ^, D; `* m# ?9 ?the validity of generated hybrid fifitting model, 6 prediction problems are chosen for experiment studies. The # d/ {. L" d! F: r3 |% lexperimental results show that the proposed method is superior to 7 typical methods in terms of the prediction0 p6 G+ V$ q( z! r1 U
accuracy and interpretability 6 V2 ]7 J1 M* B' c / P# j0 P4 y& R0 i 7 V1 }* p6 U9 W: r2 @3 q3 N( E- b& p8 j
