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
/ J7 @; L% j ]6 k
Data Fitting Method
- g, r) ?, k* }4 _* R0 T ' ]% n8 A* S9 g; n) p/ D* s7 I1 ]- l( r- B" T' @1 F
- l+ w; {* S8 z! w" d
Data fifitting is the process of constructing a curve, or a set of mathematical functions, that has% T' \- Y) q. Z1 ~* b# \
the best fifit to a series of data points. Different with constructing a fifitting model from same type of function,4 ?) |' ?) |" c
such as the polynomial model, we notice that a hybrid fifitting model with multiple types of function may have ; ?* J7 g9 T7 S; Na better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid- l( T" K3 t% D/ q* x' X
fifitting model depends on a reasonable combination of multiple functions and a set of effective parameters.2 o7 Z. t" d8 g9 ?- w
That is a high-dimensional multi-objective optimization problem. This paper proposes a novel data fifitting % w' s' y8 w+ n5 o5 _/ {. Wmodel construction approach. In this approach, the model is expressed by an improved tree coding expression 5 m! A8 \ W. Q) H9 W. P r4 hand constructed through an evolution search process driven by the genetic programming. In order to verify3 y7 H( W+ n2 q; C- D* T% x& `) q% V
the validity of generated hybrid fifitting model, 6 prediction problems are chosen for experiment studies. The! v8 m9 j3 k7 o' \8 ]
experimental results show that the proposed method is superior to 7 typical methods in terms of the prediction - _% q. U& I9 Oaccuracy and interpretability 5 m+ R+ T) I( p; [3 m- H6 i" t1 x f6 G% D! O* x' Y# v
; |# I8 n& A K8 }6 L8 J" Q5 X! V7 t' K" y- {+ L p* }1 R5 Z8 r
+ H/ C' y, @* ^2 {( d: F
