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
* C4 s* m/ w1 ?3 h' g4 B( G
Data Fitting Method
+ |4 r0 Z" s7 G S$ K" E- o
5 p- S# q3 U' N9 _$ g% K ; m' Y& ?6 o- ~" D' d; S/ s3 s3 j R, r" X/ I9 R) r8 R: O. R2 x
Data fifitting is the process of constructing a curve, or a set of mathematical functions, that has 2 m8 [0 B3 C! m& Z- _the best fifit to a series of data points. Different with constructing a fifitting model from same type of function,7 [5 i( f2 @3 S5 \( V' V
such as the polynomial model, we notice that a hybrid fifitting model with multiple types of function may have 3 P: c( \, Q: L: Ja better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid+ K# ]. T6 q, }, j
fifitting model depends on a reasonable combination of multiple functions and a set of effective parameters." C! d7 f# M/ q4 o+ d& M9 o
That is a high-dimensional multi-objective optimization problem. This paper proposes a novel data fifitting : p. D2 o8 F3 L! G0 z3 B- Zmodel construction approach. In this approach, the model is expressed by an improved tree coding expression ; x# [$ a& k) j* Uand constructed through an evolution search process driven by the genetic programming. In order to verify0 w9 y5 _, i2 F2 z6 j* Q/ A1 g
the validity of generated hybrid fifitting model, 6 prediction problems are chosen for experiment studies. The ' I+ e9 D. ?# W A# C% v+ Lexperimental results show that the proposed method is superior to 7 typical methods in terms of the prediction * |! K# x7 |9 ]! ]* d; }0 Uaccuracy and interpretability. $ v6 ^( U( M4 x) J' s% n7 @
1 I$ D# ?! \0 h l3 H w1 U" _5 o& z9 t4 l! P# b# p: u
: Y$ ^7 H7 o& J9 ]5 q& I. @2 F! P2 h5 @9 L
1 x) [; r! a N) K
