A Genetic Programming-Driven Data Fitting Method

杨利霞

A Genetic Programming-Driven

6 B# b3 y: d2 C# Y7 ^

Data Fitting Method

4 f% ]( d1 C; s0 \; _# _7 W2 N  e% _

0 s  n+ [: F* t" y) E, M) Q
Data fifitting is the process of constructing a curve, or a set of mathematical functions, that has
6 S  `6 Y* |7 V; p) [1 uthe best fifit to a series of data points. Different with constructing a fifitting model from same type of function,
# ]5 [& \, L% n0 ksuch as the polynomial model, we notice that a hybrid fifitting model with multiple types of function may have
a better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid
! n8 r8 B0 B% {fifitting model depends on a reasonable combination of multiple functions and a set of effective parameters.
That is a high-dimensional multi-objective optimization problem. This paper proposes a novel data fifitting
7 I  t7 z* L3 q+ X4 M; }/ K8 |model construction approach. In this approach, the model is expressed by an improved tree coding expression
and constructed through an evolution search process driven by the genetic programming. In order to verify
* V* N* V% H; \  Z$ y! W7 Y, _! {the validity of generated hybrid fifitting model, 6 prediction problems are chosen for experiment studies. The
" ]& Y7 u5 d, t9 b) N* }& ^1 Nexperimental results show that the proposed method is superior to 7 typical methods in terms of the prediction
& h3 F9 T' o7 {9 ^accuracy and interpretability.
0 S2 h1 @% k& B4 J$ ^8 G