Databases Reference
In-Depth Information
The correlation matrix
After viewing the summary statistics for the Iris dataset, evaluate the relation-
ships between attributes in the data. In VisMiner, a good starting point is the
correlation matrix.
To open the Iris dataset in a correlation matrix viewer, drag the dataset icon
up to an available display and drop. A context menu will open, listing all of
the available viewers for the dataset.
Select “Correlation Matrix”.
The correlation matrix (Figure 2.7) visually presents the degree of correlation
between each possible pairing of attributes in the dataset. Direct correlations are
represented as a shade of blue. The more saturated the blue the stronger the
correlation. Inverse correlations are represented as a shade of red, again with
saturation indicating the strength of the correlation. Between pairings of
numeric attributes, the coefficient of correlation is encoded using the blue
or red colors. Between pairings of nominal and numeric attributes the
Figure 2.7
Correlation Matrix
