Skip to main content

%DeepSee.PlugIn.PMML

Class %DeepSee.PlugIn.PMML Extends %DeepSee.KPIPlugIn [ System = 4 ]

This PlugIn implementation executes a predictive model defined in a %DeepSee.PMML.Definition class for every fact in the cell for which the plugin is calculated, and aggregates the predictions. By default, the plugin returns a PredictedValue property containing the aggregated prediction. Additional properties (not listed in KPI) are provided for all elements produced by the PMML model.

Note: The plugin expects an extension element mapping the DeepSee fact to the PMML's data fields in order to be able to execute the model.

Filters (configurable):

  • %cube: The cube this KPI is executed on
  • aggregate (default = "average"): How to aggregate the predicted values over different facts.
    Available values are sum, average, max, min, maxFreqValue, maxProbValue, minFreqValue and minProbValue.
  • PMML: Full class name of the %DeepSee.PMML.Definition containing the model to be executed.
  • ModelName: Name of the model in the class specified through PMML that needs to be executed. If left blank, the first model in the PMML definition will be executed.
  • NullValues (default = "ignore"): Whether or not to include null predictions when aggregating results. Available values are ignore and count.

Parameters​

BASECUBE​

Parameter BASECUBE = "*";

LISTINGSOURCE​

Parameter LISTINGSOURCE = "FactTable";

PLUGINTYPE​

Parameter PLUGINTYPE = "Aggregate";

LISTINGFIELDS​

Parameter LISTINGFIELDS = "ID";

XDatas​

KPI​

XData KPI [ XMLNamespace = "http://www.intersystems.com/deepsee/kpi" ]

This XData definition defines the KPI.

Methods​

GetCubeName​

Method GetCubeName() As %String [ Internal ]

%OnGetMDX​

Method %OnGetMDX(ByRef pMDX As %String) As %Status

Get the base query for this plug-in.

%OnCompute​

Method %OnCompute(pSQLRS As %SQL.StatementResult, pFactCount As %Integer) As %Status

Compute the median of values within the result set.