Logistic regression analysis: An application on health data

Many studies conducted in today's conditions are based on the ability to obtain forward results using available data. This is the aim of many studies in the field of social and health sciences. The result to be obtained has increased the importance of the statistical method used. In this thesis, logistic regression analysis, which is mostly used as a preferred method in the studies carried out using statistical analysis for predictive purposes, was discussed in detail. In the first part of the study, some information on logistic regression analysis was given and then binary logistic regression analysis and the reasons why this method has been preferred were emphasized. Parameter estimation methods in univariate and multivariable logistic regression models, significance tests of coefficients, interpretation of model coefficients, multiple regression analysis, and model selection methods were explained in detail. In the following sections, a literature review of the studies using logistic regression analysis was given. Then, by using full and forward variable addition methods, logistic regression analysis was performed between the operated patients diagnosed with uterine cancer and selected from the hospital database by retrospectively screening and the patients in the control group operated but not diagnosed with cancer. Afterwards, the values found using the backward variable elimination method on the same data set were compared. Significant results for diagnosis of the disease were achieved with the obtained data.