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
6 i' u* f9 z: P
Data Fitting Method
5 U! s1 O- }+ n- i3 L3 y! \6 k: _6 r+ S; [
/ N( J5 B& H1 T$ ?9 e: i, n8 `5 e6 d O! K9 n# x! W1 _
Data fifitting is the process of constructing a curve, or a set of mathematical functions, that has9 I1 x' P/ i) c$ @
the best fifit to a series of data points. Different with constructing a fifitting model from same type of function,# L3 L1 J$ r# Q0 h* M% Z
such as the polynomial model, we notice that a hybrid fifitting model with multiple types of function may have . E; E) w" O" y6 |4 d" @ U8 ia better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid `! ~. {* C7 f2 s; k2 G
fifitting model depends on a reasonable combination of multiple functions and a set of effective parameters. E( i4 @2 G9 n% L: Z3 }( K
That is a high-dimensional multi-objective optimization problem. This paper proposes a novel data fifitting 9 Q( |# J9 Q8 n) I* [model construction approach. In this approach, the model is expressed by an improved tree coding expression 7 N1 e b; O. `- F# W$ Land constructed through an evolution search process driven by the genetic programming. In order to verify + U) u+ @# x! f0 |the validity of generated hybrid fifitting model, 6 prediction problems are chosen for experiment studies. The* p% D6 f) b% T8 N9 u( E5 t
experimental results show that the proposed method is superior to 7 typical methods in terms of the prediction 5 i4 A- B3 }: [; ~. kaccuracy and interpretability. 6 }; x5 {0 `3 c2 p; g) Q
& `/ b& M- F' }, `$ o4 l # K% D/ w) u; |- A( [7 w; D/ g i) L: k5 p
- f, l2 C: @1 o% r; X2 B& g
