Selection of data analysis techniques for data mining applications

Multivariate statistical techniques are used to analyze complex data sets with marry independent and dependent variables. The dataset may be analyzed for relationships among variables based on correlation. significance of group differences based on variance and covariance, prediction of group membership, and prediction of empirical or theoretical structure of the data. The choice among the available multivariate analysis techniques for each of these research questions is based on the nature of the variables, the number of independent and dependent variables and if the independent variables can be considered as covariates. This paper describes a software tool that can assist researchers in selecting the appropriate data analysis technique based on the research needs of the data. The data analyses techniques discussed in this paper are discriminant function analysis, multi-way frequency analysis and logistic regression. The structure underlying a dataset is based on multivariate approaches such as principal components analysis, factor analysis and structural equation modeling. The paper illustrates the software tool on the Fisher's Iris data set.