Scaling in Data Editor
Scaling affects the results of classification analysis, since it determines what correlations are important. It is advisable to test different scaling options. The Data editor offers two ways of scaling data:
Scaling in the Data editor is always performed on columns, and it is highly recommended that genes are arranged in columns and samples in rows. To scale rows (samples) instead, transpose the data before scaling. After the data is scaled, press the Transpose button again to get the data on its origninal form. The scaling in the Data editor is irreversible (though, you can always press Undo). If you want to test different scaling options, it might be better to create separate data sets in the Data manager, each containing the same data file, but with different scaling options chosen in the Data manager.
For further examples and information about mean centering and autoscaling, see the tutorials Scaling during pre-processing and Scaling during processing at www.multid.se