Tucker-1 Plots
Previous Top Next
Two types of plots are generated for the Tucker-1 method. The common scores plot, the first
type, displays how the samples relate to each other based on the assessors’ evaluations. In
the correlation loadings plot, the other type, each point displays the attribute of a specific
assessor.
The more noise an attribute of an assessor contains, the closer the point will appear in the
middle. The more structured information an attribute contains, the closer it will appear to the
outer circle (100 % explained variance for that attribute).
In both plots the first two principal components (PC) are shown. The buttons with arrows are
used to change the PC on the x- and y-axes. At most 4 PCs are possible.
It is possible to highlight either one specific assessor or one specific attribute in the plot.
In the tree control all assessors and attributes are listed, together with “Common Scores” (on
the top of the control tree). By selecting a specific assessor, all correlation loadings for this
assessor are plotted. Similarly, for a specific attribute.
By default no Variable standardization of the data takes place. Alternatively, data may be
standardized by checking the respective radio button.
Variable standardization
For the Tucker-1 analysis, data may be standardized the following way:
xnew = ( xnew - xnew ) / STD(x)
This is also known as autoscaling. By default the value is None.