autoclassifiernode properties

Auto Classifier node iconThe Auto Classifier node creates and compares a number of different models for binary outcomes (yes or no, churn or do not churn, and so on), allowing you to choose the best approach for a given analysis. A number of modeling algorithms are supported, making it possible to select the methods you want to use, the specific options for each, and the criteria for comparing the results. The node generates a set of models based on the specified options and ranks the best candidates according to the criteria you specify.


node = stream.create("autoclassifier", "My node")
node.setPropertyValue("ranking_measure", "Accuracy")
node.setPropertyValue("ranking_dataset", "Training")
node.setPropertyValue("enable_accuracy_limit", True)
node.setPropertyValue("accuracy_limit", 0.9)
node.setPropertyValue("calculate_variable_importance", True)
node.setPropertyValue("use_costs", True)
node.setPropertyValue("svm", False)
Table 1. autoclassifiernode properties
autoclassifiernode Properties Values Property description
target field For flag targets, the Auto Classifier node requires a single target and one or more input fields. Weight and frequency fields can also be specified. See Common modeling node properties for more information.
ranking_measure Accuracy
ranking_dataset Training
number_of_models integer Number of models to include in the model nugget. Specify an integer between 1 and 100.
calculate_variable_importance flag  
enable_accuracy_limit flag  
accuracy_limit integer Integer between 0 and 100.
enable_area_under_curve_limit flag  
area_under_curve_limit number Real number between 0.0 and 1.0.
enable_profit_limit flag  
profit_limit number Integer greater than 0.
enable_lift_limit flag  
lift_limit number Real number greater than 1.0.
enable_number_of_variables_limit flag  
number_of_variables_limit number Integer greater than 0.
use_fixed_cost flag  
fixed_cost number Real number greater than 0.0.
variable_cost field  
use_fixed_revenue flag  
fixed_revenue number Real number greater than 0.0.
variable_revenue field  
use_fixed_weight flag  
fixed_weight number Real number greater than 0.0
variable_weight field  
lift_percentile number Integer between 0 and 100.
enable_model_build_time_limit flag  
model_build_time_limit number Integer set to the number of minutes to limit the time taken to build each individual model.
enable_stop_after_time_limit flag  
stop_after_time_limit number Real number set to the number of hours to limit the overall elapsed time for an auto classifier run.
enable_stop_after_valid_model_produced flag  
use_costs flag  
<algorithm> flag Enables or disables the use of a specific algorithm.
<algorithm>.<property> string Sets a property value for a specific algorithm. See Setting algorithm properties for more information.
use_cross_validation field Fields added to this list can take either the condition or prediction role in rules that are generated by the model. This is on a rule by rule basis, so a field might be a condition in one rule and a prediction in another.
number_of_folds integer N fold parameter for cross validation, with range from 3 to 10.
set_random_seed boolean Setting a random seed allows you to replicate analyses. Specify an integer or click Generate, which will create a pseudo-random integer between 1 and 2147483647, inclusive. By default, analyses are replicated with seed 229176228.
random_seed integer Random seed
stop_if_valid_model boolean
filter_individual_model_output boolean Removes from the output all of the additional fields generated by the individual models that feed into the Ensemble node. Select this option if you're interested only in the combined score from all of the input models. Ensure that this option is deselected if, for example, you want to use an Analysis node or Evaluation node to compare the accuracy of the combined score with that of each of the individual input models
set_ensemble_method "Voting"
Ensemble method for set targets.
set_voting_tie_selection "Random"
If voting is tied, select value randomly or by using highest confidence.
flag_ensemble_method "Voting"
Ensemble method for flag targets.
flag_voting_tie_selection "Random"
If voting is tied, select the value randomly, with highest confidence, or with raw propensity.