Investigation of logistics regression models in data mining

In data mining, many methods are used to reveal cause-effect relationships between variables and to model these relationships. Regression Models constitute an important part of these methods used in Data Mining. Logistic regression models, on the other hand, are widely used when the dependent variable is a categorical variable with two or more groups. In this study, Logistic Regression Analysis; By considering Binary, Multinomial, and Multiordinal Logistic Regression Models, were analyzed with open source R programming language and the results were interpreted. Before model estimation; The basic "Data Mining Pre-Analysis Preparation Process" was completed with the cleaning, integration, reduction, and transformation of the data. It was tested whether the Logistic Regression Model, which was determined according to the data type, provided the assumptions. Binary, Multinomial and Multiordinal models; Models with one independent variable (continuous, two-category qualitative, and more than two-category qualitative) and at least two independent variables (continuous, two-category qualitative, and more than two-categorical qualitative) models, that is, all possible Logistic Regression Models, were estimated and interpreted in detail. The results that can be visualized are presented visually, thus making the analysis results more understandable. During the analysis process, some results that could not be reached with the codes in the R packages were obtained by writing new codes. Briefly, the main purpose of this study is; In Logistic Regression Analysis applications, choose the suitable model for the data type, apply the Data Mining Pre-Analysis Preparation Processes based on the theory of the selected model, estimating the model in the R programming language and interpreting the results.