Last updated: Jan 17, 2024
The Ensemble node combines two or more model nuggets to obtain more accurate
predictions than can be gained from any one model.
Example
# Create and configure an Ensemble node
node = stream.create("ensemble", "My node")
node.setPropertyValue("ensemble_target_field", "response")
node.setPropertyValue("filter_individual_model_output", False)
node.setPropertyValue("flag_ensemble_method", "ConfidenceWeightedVoting")
node.setPropertyValue("flag_voting_tie_selection", "HighestConfidence")
ensemblenode properties |
Data type | Property description |
|---|---|---|
ensemble_target_field
|
field | Specifies the target field for all models used in the ensemble. |
filter_individual_model_output
|
flag | Specifies whether scoring results from individual models should be suppressed. |
flag_ensemble_method
|
Voting
ConfidenceWeightedVoting
RawPropensityWeightedVoting
AdjustedPropensityWeightedVoting
HighestConfidence
AverageRawPropensity
AverageAdjustedPropensity
|
Specifies the method used to determine the ensemble score. This setting applies only if the selected target is a flag field. |
set_ensemble_method
|
Voting
ConfidenceWeightedVoting
HighestConfidence
|
Specifies the method used to determine the ensemble score. This setting applies only if the selected target is a nominal field. |
flag_voting_tie_selection
|
Random
HighestConfidence
RawPropensity
AdjustedPropensity
|
If a voting method is selected, specifies how ties are resolved. This setting applies only if the selected target is a flag field. |
set_voting_tie_selection
|
Random
HighestConfidence
|
If a voting method is selected, specifies how ties are resolved. This setting applies only if the selected target is a nominal field. |
calculate_standard_error
|
flag | If the target field is continuous, a standard error calculation is run by default to calculate the difference between the measured or estimated values and the true values; and to show how close those estimates matched. |