Transformation functions
The machine learning API includes a set of UDx functions that transform the columns of each input row to one or more corresponding output columns.
The machine learning API includes a set of UDx functions that transform the columns of each input row to one or more corresponding output columns. These transformations follow rules that are defined in models that were created earlier. For example,
APPLY_SVD
uses an SVD model to transform input data.
Unless otherwise indicated, these functions require the following privileges for non-superusers:
-
USAGE privileges on the model
-
SELECT privileges on the input relation
In general, given an invalid input row, the return value for these functions is NULL.
In this section
- APPLY_BISECTING_KMEANS
- APPLY_IFOREST
- APPLY_INVERSE_PCA
- APPLY_INVERSE_SVD
- APPLY_KMEANS
- APPLY_KPROTOTYPES
- APPLY_NORMALIZE
- APPLY_ONE_HOT_ENCODER
- APPLY_PCA
- APPLY_SVD
- PREDICT_ARIMA
- PREDICT_AUTOREGRESSOR
- PREDICT_LINEAR_REG
- PREDICT_LOGISTIC_REG
- PREDICT_MOVING_AVERAGE
- PREDICT_NAIVE_BAYES
- PREDICT_NAIVE_BAYES_CLASSES
- PREDICT_PMML
- PREDICT_POISSON_REG
- PREDICT_RF_CLASSIFIER
- PREDICT_RF_CLASSIFIER_CLASSES
- PREDICT_RF_REGRESSOR
- PREDICT_SVM_CLASSIFIER
- PREDICT_SVM_REGRESSOR
- PREDICT_TENSORFLOW
- PREDICT_XGB_CLASSIFIER
- PREDICT_XGB_CLASSIFIER_CLASSES
- PREDICT_XGB_REGRESSOR
- REVERSE_NORMALIZE