|Top Previous Next|
The display options are selected from the panel to the left of the graph. This panel can be shrunk to the left to make the plot as large as possible. Above the plot there is a toolbar of commonly-used graphics editing tools.
The label radio box on the left under Plot options is used to select labelling for the samples.
Select: Title to display on the plot the names of the samples.
Number to show the sequential number of each sample in the working data set.
No labels to remove labels from the plot.
The Plot radio box, on the left under Plot options, allows sample scores, the eigenvectors (one for each species or row in the working data ) or both to be plotted. If both are selected then the biplot uses different axes to plot the scores and eigenvectors. The axes for the scores are displayed at the bottom and on the left.
The axes displayed in the plot are selected using the Axes to plot tab. The default is axis 1 and axis 2, which will display the relative positions of the samples with respect to the two largest components. A 3D plot is produced if a z variable is selected.
Use the Find point tab to locate the position of an individual point on the chart. In the top pane of the tab is a list of the sample groups; click on a group and the individual samples in that group are shown in the lower pane. Double-click and hold down on the sample you wish to locate; the corresponding point on the chart will be marked by a red cross-hairs symbol:
Setting a perimeter
It is possible to display the perimeter of each predefined group. Below is an example output for Fisher's iris data showing the grouping of the 3 species, with each group outlined:
If you have performed a PCA on an ungrouped data set and, on inspection of the plot, you decide that a group exists that you wish to emphasise, e.g. for publication, then it will be necessary to scrutinise the plot to identify and take note of all the samples that you wish to encircle. Click on the Grouping tab to show the group membership editing page, assign those samples to a group and then re-run the PCA. Set data point colours as preferred and then create the perimeter line.