A Genetic Programming-Driven Data Fitting Method

杨利霞

A Genetic Programming-Driven

Data Fitting Method

: C% ^) R; c" M' _8 b# D
& X: V  j  q0 M5 J6 l1 z
Data fifitting is the process of constructing a curve, or a set of mathematical functions, that has
9 Z, V$ Y: e& u: Z6 }3 [$ othe best fifit to a series of data points. Different with constructing a fifitting model from same type of function,
; s2 w: V5 @% J$ r  Z2 }such as the polynomial model, we notice that a hybrid fifitting model with multiple types of function may have
4 h/ K: o( O  b, ^. Aa better fifitting result. Moreover, this also shows better interpretability. However, a perfect smooth hybrid
8 j& m8 t1 k1 i6 ?" Hfifitting 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 L7 O0 E, U. e$ Pmodel construction approach. In this approach, the model is expressed by an improved tree coding expression
; c# g9 b$ X" H2 Rand constructed through an evolution search process driven by the genetic programming. In order to verify
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
accuracy and interpretability
& v  A( z% i8 C& ^
6 Y' G: R$ s* K. o6 |' R: p4 i/ Y
- D$ \6 U( {, A- M
8 M# g! O3 d) R2 k
0 B$ [& [) v& _( k, |