Outline

There are two data sets (Table E.1) that differ in the amount of correlation. Each data set has four predictors, a1–a4 and b1–b4, and there is a class variable that is common to both. The predictors a1–a4 are uncorrelated with each other while there are some correlations within data set B.

The data

Descriptive statistics and relationships between variables

Both data sets show a trend of increasing means across both classes that is matched by a similar increase in the standard errors (Table E.2). Table E.2 and Figure E.3 clearly show the different correlation structures within the two data sets. There are no significant between-predictor correlations in data set A. However, in data set B b3 and b4 are highly correlated, although all other between-predictor correlations are insignificant.