ai.h2o.mojos.runtime

Interface MojoPipelineFactory

public interface MojoPipelineFactory

Factory class for loading MOJO pipeline objects.

“Pipeline” means an entity for generic data transformation. It takes some input data, and produces one or more outputs. The outputs may be arbitrary transforms of the inputs, including new generated features, predictions from some machine learning models, statistical estimates of accuracy, etc.

“Mojo” (Model ObJect Optimized) is the H2O’s format for saving the models into a file.

Method Summary

Modifier and Type	Method and Description
java.lang.Object	loadFrom(MojoReaderBackend backend) Advanced way of constructing instances of MojoPipeline by supplying a custom mojo reader backend.
java.lang.Object	loadFrom(java.lang.String file) A factory method for constructing MojoPipeline instances from a file.

Method Detail

loadFrom

java.lang.Object loadFrom(java.lang.String file)
                   throws java.io.IOException,
                          LicenseException

A factory method for constructing MojoPipeline instances from a file.

Parameters:

file - Name of the model’s .mojo file.

Returns:

a loaded pipeline

Throws:

java.io.IOException - if file does not exist, or cannot be read, or does not represent a valid model.

LicenseException

loadFrom

java.lang.Object loadFrom(MojoReaderBackend backend)
                   throws java.io.IOException,
                          LicenseException

Advanced way of constructing instances of MojoPipeline by supplying a custom mojo reader backend. For example a FolderMojoReaderBackend can be used to read an unzipped mojo model from a folder. Alternatively, you can implement your own backend, for example to read models stored on network locations, or load them from a database, etc.

Parameters:

backend - a class implementing the MojoReaderBackend interface.

Returns:

a loaded pipeline

Throws:

java.io.IOException - if mojo reader fails to read the model, or if the model retrieved is not valid.

LicenseException