A Genetic Programming-Driven Data Fitting Method

杨利霞

A Genetic Programming-Driven

Data Fitting Method

$ Z- ]' u! t# s- J. ^
2 W# }* O5 n3 P+ n- N8 r5 _" w
) Z. S6 t- A7 k, U$ h# z! j
/ x1 X: @1 t, ~; d- m( mData fifitting is the process of constructing a curve, or a set of mathematical functions, that has
8 T* u9 c" e. t# C5 P) g; I& {the 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
a 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.
  i2 U% r7 ]* }That 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
$ }4 I$ g" F8 u3 D9 z9 }2 T) m+ Iand 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
' J% @( D$ r& i" P9 rexperimental results show that the proposed method is superior to 7 typical methods in terms of the prediction
/ X% z8 ?/ o% G/ [" Q. ~4 o& q& Raccuracy and interpretability.



6 d9 {, D9 X/ d" c
: ?- \: W  t8 a, ~2 d8 w
. S/ A1 M5 q* e3 t; M/ Z$ T8 f& d