Optimal scaling techniques and an application

In the studies conducted in the field of Social Sciences, the data consists of categorical variables. However, working with such data causes problems in providing many assumptions. Developing solutions to this problem by applying transformation to categorical data is important for many researchers. One of these solutions is Optimal Scaling or Gif Method. With Optimal Scaling, the categorical data is transformed into continuous data without disturbing the scale properties and information loss is prevented. In this thesis; Regression analysis was used to determine whether mobile phone use has an effect on sleep patterns, and if so, which variables are explained. In the research, the survey data applied to a total of 491 people were analyzed. The data set consists of 25 independent variables, all of which are categorically dependent. Applying Regression Analysis to so many independent variables at the same time will create difficulties in terms of both productivity and forecasting. The main argument of our research is to determine the most appropriate model with the least variable. Genetic Algorithms, which is an intuitive method that seeks solutions to the researcher by incorporating many optional options, are used for this purpose. Thus, an optimal result will be achieved by finding the best solution for a specific purpose function. For this purpose, ICOMP (Information Complexity) and ICOMP (IFIM), which takes into account the complexity of the model as well as the model simplicity and goodness of fit developed by Hamparsum Bozdoğan (1988), have been determined as the objective function to be used for this purpose. As a result, the necessary constraints and choices were made for the algorithm, and the regression model that could explain the dependent variable among the 25 independent variables and gave the most appropriate subset to the objective function was determined and the model was created