ANTICIPATION OF DYNAMICS AND IDENTIFICATION OF POTENTIAL FACTOR VARIABLES. A CASE STUDY

Elmira KUSHTA, Miftar RAMOSACAJ

Abstract

Based on this paper we have identified the true factor variables [X] and that response [Y] for the given system the first variable does not give a good match for any combination. This response variable is inappropriate for judging the system as a whole. Taking another variable that resulted in a good logic regression, we see that the elimination of the first 2 variables does not significantly change the fit of the regression. Again, the regression is not very good, which indicates the presence of other factors included in this study, but for the case with 5 variables, the pepper has a good deal with all the variables We note that the technique used with linear OLS and non-linear OLS can be well adapted to the logistic case. By imposing a practical balance between regression fitting goodness and the quality of the reproduction of the values for response variables, we decided to set-up a well-designed logistic model to analyze consumer behavior for a real market (district of Vlora). The steps proposed herein have worked as a data–oriented modeling methodology that reduces subjective or empirical approaches in econometric and marketing analysis.

Pages: 356 - 364