Prediction and Modeling Operator Pairings

In general, the Classification Modeling operators are followed in a workflow by a Classifier operator, and the Regression Modeling operators that predict a value are followed by a Predictor Operator.

However, a Predictor operator can also be applied to Classification models in order to return the value with the highest probability.

The few exceptions are for modeling techniques such as Collaborative Filtering, Time Series, EM Cluster, PCA, and LDA that have their own specialized Predictor operator (shown in the following table). See Prediction Operators for more information about the classifiers and predictors.

The following table summarizes which Prediction operators can follow Modeling operators (to apply the appropriate modeling algorithms).

Modeling operator Next operator Exception
ARIMA Time Series n/a Time Series Predictor
Alpine Forest Classifier or Predictor n/a
Alpine Forest Regression Predictor n/a
Association Rules n/a n/a
Collaborative Filter Trainer n/a Collaborative Filter Predictor, Collaborative Filter Recommender
Decision Tree Classifier or Predictor n/a
Elastic Net Linear Predictor
Elastic Net Logistic Classifier or Predictor
Generalized Linear Regression Models Predictor
Gradient Boosting Classification Classifier n/a
Gradient Boosting Regression Predictor n/a
K-Means Classifier* n/a
Linear Regression Predictor n/a
Logistic Regression Classifier or Predictor n/a
Naive Bayes Classifier or Predictor n/a
Neural Network Classifier or Predictor n/a
PCA n/a PCA Apply*
SVM Classification Classifier or Predictor n/a

*PCA Apply operator and the Classifier for K-Means are used only within Hadoop workflows.