A Genetic Programming-Driven Data Fitting Method

杨利霞

A Genetic Programming-Driven

Data Fitting Method

, K# V' c3 P% e! M& L, a


Data fifitting is the process of constructing a curve, or a set of mathematical functions, that has
& z) B( u0 r4 {# F& I0 N0 ~9 A9 qthe best fifit to a series of data points. Different with constructing a fifitting model from same type of function,
such as the polynomial model, we notice that a hybrid fifitting model with multiple types of function may have
5 d: I% _6 V/ W$ o  za better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid
fifitting model depends on a reasonable combination of multiple functions and a set of effective parameters.
4 n! _; f0 ^3 m3 j% V( MThat is a high-dimensional multi-objective optimization problem. This paper proposes a novel data fifitting
model construction approach. In this approach, the model is expressed by an improved tree coding expression
# t. |2 m- W" f2 G" R! ^and constructed through an evolution search process driven by the genetic programming. In order to verify
4 n0 X' D9 I' `4 l* J! L: Athe validity of generated hybrid fifitting model, 6 prediction problems are chosen for experiment studies. The
/ ?/ u, }/ S* h. [experimental results show that the proposed method is superior to 7 typical methods in terms of the prediction
3 t9 G) G% I0 r) R% Caccuracy and interpretability.
& v, _/ n/ T$ X5 a( o
$ w8 \: }! u5 W
% `, \* Y0 D( T* a5 `2 U3 ?4 D
* ], I  r7 G" j3 T

2 k7 }4 B! P+ q5 k6 y5 x