The prediction table is used for classification of unknown observations with a PLS-DA model. This table can only be created when at least one observation is set as a test set. The prediction table is created by dragging the PLS-DA model from the Data Tree onto the Table Area.

The prediction table specifies how many of the test set observations that are predicted as a certain class.

The default low cut-off for an observation to be considered to belong to a class is 0.5 and high cut-off is 1.5. The cut-off can be changed by right-clicking on the prediction table and choose properties from the menu.

If the low cut-off limit is increased, the number of observations belonging to a certain class can decrease. A cut-off limit of 0.9 has been used in the example below and this has resulted in that seventeen observations is not belonging to any of the two classes.