Comparison of Logistic Regression and Discriminate Analysis in Recognition of the Factors Affecting on the Distribution of Quercus Libanii of Armardeh Forests at Baneh, Kurdistan Province

The study site, Armardeh forest, is located in Baneh, Kurdistan Province. The study area covers approximately 17000 ha. In this study, logistic regression and discriminant analysis methods were used and their performances were compared. Two random sampling grids were overlaid on the presence and absence area of Quercus libani type as dependent variables. Physiographic factors including slope, aspect, elevation and distance from drains were extracted for each sampling site and considered as independent variables. The results showed that the area under the ROC curve (0.746) for logistic regression method is higher than the discriminant analysis with 0.502. Furthermore, the overall accuracy of the classification of the logistic regression was higher than discriminate analysis. The results of this study can be used in management and restoration of these forests.