rxSerializeModel: RevoScaleR Model Serialization and Unserialization

Article
07/15/2019

Description

Serialize a RevoScaleR/MicrosoftML model in raw format to enable saving the model. This allows model to be loaded into SQL Server for real-time scoring.

Usage

  rxSerializeModel(model, metadata = NULL, realtimeScoringOnly = FALSE, ...)

  rxUnserializeModel(serializedModel, ...)

Arguments

`model`

RevoScaleR/MicrosoftML model to be serialized

`metadata`

Arbitrary metadata of raw type to be stored with the serialized model. Metadata will be returned when unserialized.

`realtimeScoringOnly`

Drops fields not required for real-time scoring. NOTE: Setting this flag could reduce the model size but rxUnserializeModel can no longer retrieve the RevoScaleR model

`serializedModel`

Serialized model to be unserialized

Details

rxSerializeModel converts models into raw bytes to allow them to be saved and used for real-time scoring.

The following is the list of models that are currently supported in real-time scoring:

* RevoScaleR

* rxLogit

* rxLinMod

* rxBTrees

* rxDTree

* rxDForest

* MicrosoftML

* rxFastTrees

* rxFastForest

* rxLogisticRegression

* rxOneClassSvm

* rxNeuralNet

* rxFastLinear

RevoScaleR models containing R transformations or transform based formula (e.g "A ~ log(B)") are not supported in real-time scoring. Such transforms to input data may need to be handled before calling real-time scoring.

rxUnserializeModel method is used to retrieve the original R model object and metadata from the serialized raw model.

Value

rxSerializeModel returns a serialized model.

rxUnserializeModel returns original R model object. If metadata is also present returns a list containing the original model object and metadata.

Author(s)

Microsoft Corporation Microsoft Technical Support

Examples


 myIris <- iris
 myIris[1:5,]
 form <- Sepal.Length ~ Sepal.Width + Petal.Length + Petal.Width + Species
 irisLinMod <- rxLinMod(form, data = myIris)

 # Serialize model for scoring
 serializedModel <- rxSerializeModel(irisLinMod)

 # Save model to file or SQL Server (use rxWriteObject)
 # serialized model can now be used for real-time scoring

 unserializedModel <- rxUnserializeModel(serializedModel)

Partager via