标题: A Genetic Programming-Driven Data Fitting Method [打印本页] 作者: 杨利霞 时间: 2020-11-10 16:01 标题: A Genetic Programming-Driven Data Fitting Method
A Genetic Programming-Driven
) s3 F$ P; o8 V3 D" T! U% s
Data Fitting Method
7 Q1 l4 g; J( Y: g j u8 O
7 y6 D: L) D" P1 {) T9 ~
: N9 u- ^( G# p" _/ j. L7 R% \# |7 T+ a& ]2 g$ X
Data fifitting is the process of constructing a curve, or a set of mathematical functions, that has& ] V0 M: Z+ t
the best fifit to a series of data points. Different with constructing a fifitting model from same type of function, / k' C) @1 v. l# g hsuch as the polynomial model, we notice that a hybrid fifitting model with multiple types of function may have : P; V: F; F1 u& y: }) la better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid # k1 s( ` e$ q& ]" y" F% K! Xfifitting model depends on a reasonable combination of multiple functions and a set of effective parameters. : N9 B# f }: ^1 NThat is a high-dimensional multi-objective optimization problem. This paper proposes a novel data fifitting v4 b* S, i; |1 I$ kmodel construction approach. In this approach, the model is expressed by an improved tree coding expression4 i( p7 t& U$ @5 Q u. M5 h9 @7 u R, N
and constructed through an evolution search process driven by the genetic programming. In order to verify0 d* f* g3 Y+ H. y6 ^2 J d
the validity of generated hybrid fifitting model, 6 prediction problems are chosen for experiment studies. The : l/ [1 d; h. Q9 ^* v! Uexperimental results show that the proposed method is superior to 7 typical methods in terms of the prediction : C5 a" U( W, f& O/ s7 ]# |% G) w! haccuracy and interpretability. / j r' s) M2 `
9 z5 \: Y- L( W5 e* P, d3 G. c1 e$ h. W# i3 Z* J2 j/ C6 B5 P
9 O0 Q; g, U$ O; V$ q: r. O4 s* |8 ~1 c0 v2 N( l
" [( ~% l8 I7 n- D6 w
