Univariate Statistics

Abstract

Univariate analysis of data means to deal with each variable separately. How to choose the method to examine the variables highly depends upon the so-called “scale type” of each variable. Therefore, it is important to assign the correct scale to the variables in the first place, before starting the analysis. In this chapter, the reader is familiarized with the concept of discrete and continuous variables. A procedure for determining the correct scale type is outlined in detail.

Based on the decision of the scale type of a variable in the next step, the information must be extracted and represented in a well comprehensible form. To do so, values are classified or binned. The different methods offered by the IBM SPSS Modeler are explained in this chapter. Additionally, the concept of SuperNodes is being outlined in the case when normalizing variables.

After finishing this chapter, the reader is able to …

• 1. Explain in detail the necessity and the characteristics of different scale types as well as to keep hold of the big picture

• 2. Create diagrams of frequency distributions to assess the shape and to determine outliers after assigning the correct scale type to a variable

• 3. Describe the necessity to use reclassification or binning procedures to determine frequencies of values in specific intervals and finally

• 4. Use SuperNodes to transform variables and compact streams

So the successful reader is familiar with the statistical theory of assigning correct scale types to variables and after that to select and apply correct methods to show, determine, and assess frequency distributions.