Learn and test data file format
Previous  Top  Next

The learn and test data files have got to contain an ASCII data matrix. Each row corresponds to a data tuple and each column corresponds to a variable. The number of rows and columns is detected automatically. The learn data file has got to contain an output column, whereas this is not necessary for the test data file. However, the input and - if present - the output columns in the test data file must be ordered in the same way as in the learn data file.

The
minimum number of tuples/rows in the learn data file is 10. The test data file must have at least 2 tuples/rows. Actually it should be possible to have test data files with just one data tuple, but at the moment, there are technical reasons that require at least 2.

It is assumed, that the symbols of a nominal input are encoded as whole numbers starting from 1.
For example the three symbols a, b and c of a nominal input should be encoded as the values 1, 2 and 3.


Note: The data file must be in the Windows ASCII format, i.e. lines must be ended by a pair of carriage return and linefeed. If you have files in the UNIX ASCII format, you previously have got to convert them!