| Induce Single Node | The Stop button | |
This function will generate a tree automatically by rule induction. As this process involves execution of SQL statements against the 'Data Source', it may take some considerable time on very large or complex data sets.
The Minimum examples in a branch parameter is the number of data records below which the induction process is not to split a tree branch further. This is a useful defence to stop the tree fitting the noise in the data. Normally this figure is set depending on the total number of records and the level of noise in the data.
For discrete outcomes set the Maximum significance level of branch to 0.5% as a guideline. For a continuous outcome you alternatively specify the F-test Percentiles cut off in place of this. This is a forward pruning parameter, with a guideline value of 1.0%.
When the induction process starts you will see a status dialog displayed, giving an indication of progress. However, as it cannot predict how large or complex the resulting tree will be, there is no way of estimating how long the process might take.