17.3.6 Classification Versus Regression on Supervised EdgeWise Models

When predicting a property, the loss function defines if the model will perform classification tasks or regression tasks.

For classification tasks, the Supervised EdgeWise model will infer labels. Even if this property is a number, the model will assign one label for each value found and classify on it. The possible losses for classification tasks are softmax cross entropy, sigmoid cross entropy, and DevNet loss.

For regression tasks, the Supervised EdgeWise model will infer values for the property. The loss for regression tasks is the MSE loss.

It is possible to select different loss functions for the supervised model by providing a LossFunction object.

• JShell
• Java
• Python

opg4j> import oracle.pgx.config.mllib.loss.LossFunctions;
opg4j> var model = analyst.supervisedEdgeWiseModelBuilder().
         setVertexInputPropertyNames("vertex_features").
         setEdgeInputPropertyNames("edge_features").
         setEdgeTargetPropertyName("labels").
         setLossFunction(LossFunctions.MSE_LOSS).
         build()

import oracle.pgx.config.mllib.loss.LossFunctions;

SupervisedEdgeWiseModel model = analyst.supervisedEdgeWiseModelBuilder()
    .setVertexInputPropertyNames("vertex_features")
    .setEdgeInputPropertyNames("edge_features")
    .setEdgeTargetPropertyName("labels")
    .setLossFunction(LossFunctions.MSE_LOSS)
    .build();

from pypgx.api.mllib import MSELoss

params = dict(edge_target_property_name="labels",
              vertex_input_property_names=["vertex_features"],
              edge_input_property_names=["edge_features"],
              loss_fn=MSELoss())

model = analyst.supervised_edgewise_builder(**params)