استخدام الاساليب اللامعلمية لتقدير الدالة التمييزية

Some assumptions must available in random variables to allow us to use the parametric methods in discriminate function estimation. One of these assumptions is that the distribution of the variables are normal distribution ,but sometimes this is not achievement so we must seek for an alternative methods to estimate the discriminant function ,one of these methods is the nonparametric estimation procedure which called Kernel estimation or Kernel method which depends on data to get the estimation. In this paper we use Kernel procedure with three functions, Epanchinikov, Gaussian & Sine functions and compare results with the parametric method by using misclassification as criterion of comparing.