CARMA Node Model Options
Model name. You can generate the model name automatically based on the target or ID field (or model type in cases where no such field is specified) or specify a custom name.
Minimum rule support (%). You can specify a support criterion. Rule support refers to the proportion of IDs in the training data that contain the entire rule. (Note that this definition of support differs from antecedent support used in the Apriori nodes.) If you want to focus on more common rules, increase this setting.
Minimum rule confidence (%). You can specify a
confidence criterion for keeping rules in the rule set. Confidence refers to
the percentage of IDs where a correct prediction is made (out of all IDs for which the rule makes a
prediction). It is calculated as the number of IDs for which the entire rule is found divided by the
number of IDs for which the antecedents are found, based on the training data. Rules with lower
confidence than the specified criterion are discarded. If you are getting uninteresting or too many
rules, try increasing this setting. If you are getting too few rules, try decreasing this setting.
Note: If necessary, you can highlight the value and type in your own value. Be aware that if you
reduce the confidence value below 1.0, in addition to the process requiring a lot of free memory,
you might find that the rules take an extremely long time to build.
Maximum rule size. You can set the maximum number of
distinct item sets (as opposed to items) in a rule. If the rules
of interest are relatively short, you can decrease this setting to speed up building the rule
set.
Note: The CARMA model building node ignores empty records when building a model if the field type is
a flag, whereas the Apriori model building node includes empty records. Empty records are records
where all the fields used in the model build have a false value.