Discovery Options - Batch Experiments
[General Discovery Options] [Single
Experiment Options]
The discovery node's batch experiment mode performs multiple runs of
the simulated annealing search engine that can search for different
types of rules. Sample batches may describe multiple classes of
records or look for rules with varying accuracy or coverage.
-
- Target Classes
- Build rules to describe records in these classes.
One set of experiments will be run for each class.
-
- Number of Runs
- The number of runs that will be performed with each
set of parameters. You can also click on the spin control to increase
or decrease the current value.
-
- Accuracy Threshold
- Check this option to run multiple experiments using
different accuracy thresholds. The min, max and step values control
the accuracy thresholds used during the experiments. See also the tips
section for help on how to set this value.
-
- Max. No. of Fields
- Check this option to run multiple experiments using
different limitations on the field length. The min, max and step
values control the length of the rule throughout the different experiments.
- Tip: if
you limit the number of conditions that can appear in a rule, your
rules will be easier to comprehend.
-
- Reset RNG Seed
- Resets the random number stream between
experiments. Click the button to set the seed for the random number
generator. The default value is 87326471.
-
- Advanced
- Advanced options are used to fine-tune the
simulated annealing (SA) search engine. Under most circumstances the
default options should be sufficient. There are seven pages of
advanced options for the SA engine.
Notes
For definitions of accuracy, coverage and other properties, see the definitions page.
