Özet:

In this study, the homogenity and heterogenity of the data in many variables have been studied and the heterogenic variables have been determined. The divisions in variables (heterogenicity) have been shown to be equal to the components in normal karma divisions and the subgroup numbers have been determined. The K-means algorithms and the observations assigned to divisions in the variables have been determined and the data grouped. Under the assumption that each fragmentation in the variables corresponds to an accumulation in the Gauss Karma Model (GMM), the range for the possible quantities and quantities is obtained and the quantities of the model are determined depending on the quantities. The number of models related to the number of divisions is determined by genetic algorithms (GA) and the parameters are predicted by the MLE algorithm. The choice of the best model that fits the data structure between the model-based aggregation method and Gauss Karma Models is determined by information criteria such as log-likelihood, AIC and BIC.

Anahtar Kelimeler: