Comparative analysis of multivariate analysis methods

Statistical methods; the results of important scientific studies are also evaluated by statistical methods, such as the average achievement of students, traffic accident statistics, productivity in agriculture and animal husbandry, economic data, as well as frequently needed events in daily life. In the simplest sense, statistical relations are based on the effect of an independent variable on a dependent variable. However, there are multiple factors that affect the variables from natural events to health events. The lack of univariate statistical analyzes and its ability to explain limited events required the use of multivariate statistical analysis methods. As an alternative to multivariate statistical methods, the application of univariate methods consecutively can be considered. However, this will neglect the interaction between variables for many methods, and also lead to an increase in random error rates. The results of univariate hypothesis testing may not be the same with successive multivariate hypothesis testing. For example, if the variables are tested individually for the normality test, the null hypothesis is accepted and all variables can be considered to conform to the normal distribution. However, variables may not provide multivariate normality together. Multivariate statistical methods have been developed to meet the needs with the principle of cumulative progression of information. Obviously, there will be no need for a new method that will be used for the same purpose and all assumptions will be the same. Since each method will be unique according to its assumptions and indicators, multivariate statistics cannot be separated into a precise classification according to their aims and methods. In this thesis, a comparison of some of the methods used to test multivariate statistics and their assumptions has been made. It has been tested whether the variables provide individual univariate normality or not to provide multivariate normality. KEYWORDS: Multivariate Statistical Methods, Multiple Normal Distribution, Discriminant Analysis, Logistic Regression Analysis, Probit Analysis, Cluster Analysis, Factor Analysis