A Genetic Programming-Driven Data Fitting Method

杨利霞

A Genetic Programming-Driven

& C- Y" u7 f+ n

Data Fitting Method

8 W0 K; j! W8 w3 ?3 e

' O7 B" e# ]: D5 ]" }; c
Data fifitting is the process of constructing a curve, or a set of mathematical functions, that has
* g9 G5 A8 e4 m! l' `7 Cthe 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
3 f! H$ Z6 C. Ja better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid
8 y# V, C. `* C! Z; X8 [& O6 g+ j* Wfifitting model depends on a reasonable combination of multiple functions and a set of effective parameters.
$ p; s1 j0 e  r" }+ |9 JThat 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
* @( O9 W- r- eand constructed through an evolution search process driven by the genetic programming. In order to verify
& G$ R6 j- `0 Y4 C- `1 @. L0 J/ ]the validity of generated hybrid fifitting model, 6 prediction problems are chosen for experiment studies. The
experimental results show that the proposed method is superior to 7 typical methods in terms of the prediction
% b* K  p! t! `7 Haccuracy and interpretability.
9 ~7 t, Z+ @9 ]( |/ e
( |4 i0 J! x8 p% e+ _2 P6 C