Polynomial Curve Fitting with Varying Real Powers

  • B. Cetisli Suleyman Demirel University
  • H. Kalkan Suleyman Demirel University


In this study, the polynomial curve fitting is expanded with real powers by combining the genetic algorithm and the traditional least squares estimator. In general, integer values are used as the power of variables in traditional polynomials. However, the usage of real powers decreased the approximation error on polynomial curve fitting. In addition, the number of parameters is also decreased, when it is compared to the traditional polynomials. But, the number of adapted parameters of the proposed method is bigger than the parameters of traditional polynomials with same conditions. Although the improvements in polynomial curve fitting, the algorithm can be used for the positive input space to avoid the complex outputs. Ill. 5, bibl. 12, tabl. 4 (in English; abstracts in English and Lithuanian).