Link Prediction Training Pipeline

class graphdatascience.pipeline.lp_training_pipeline.LPTrainingPipeline

Represents a link prediction training pipeline. Construct an instance of this class using graphdatascience.GraphDataScience.lp_pipe().

addFeature(feature_type: str, **config: Any) → Series[Any]

Add a link feature to the pipeline.

Parameters:

• feature_type – The type of feature to add.

• **config – The configuration for the feature, this includes the node properties to use.

Returns:

The result of the query.

addLogisticRegression(**config: Any) → Series[Any]

Add a logistic regression model candidate to the pipeline.

Parameters:

**config – The configuration for the logistic regression model.

Returns:

The result of the query.

addMLP(**config: Any) → Series[Any]

Add a multi-layer perceptron model candidate to the pipeline.

Parameters:

**config – The configuration for the multi-layer perceptron model.

Returns:

The result of the query.

addNodeProperty(procedure_name: str, **config: Any) → Series[Any]

Add a node property step to the pipeline.

Parameters:

• procedure_name – The name of the procedure to use.

• **config – The configuration for the node property.

Returns:

The result of the query.

addRandomForest(**config: Any) → Series[Any]

Add a random forest model candidate to the pipeline.

Parameters:

**config – The configuration for the random forest model.

Returns:

The result of the query.

auto_tuning_config() → Series[Any]

Get the auto-tuning configuration of the pipeline.

Returns:

A Series containing the auto-tuning configuration.

configureAutoTuning(**config: Any) → Series[Any]

Configure auto-tuning for the pipeline.

Parameters:

**config – The configuration for auto-tuning.

Returns:

The result of the query.

configureSplit(**config: Any) → Series[Any]

Configure the splits for training the pipeline.

Parameters:

**config – The configuration for the splits.

Returns:

The result of the query.

creation_time() → Any

Get the creation time of the pipeline.

Returns:

The creation time of the pipeline.

drop(failIfMissing: bool = False) → Series[Any]

Drop the pipeline.

Parameters:

failIfMissing – If True, an error will be thrown if the pipeline does not exist.

Returns:

The result of the query.

exists() → bool

Check if the pipeline exists.

Returns:

True if the pipeline exists, False otherwise.

feature_steps() → DataFrame

Get the feature steps of the pipeline.

Returns:

A DataFrame containing the feature steps of the pipeline.

name() → str

Get the name of the pipeline.

Returns:

The name of the pipeline.

node_property_steps() → DataFrame

Get the node property steps of the pipeline.

Returns:

A DataFrame containing the node property steps.

parameter_space() → Series[Any]

Get the parameter space of the pipeline.

Returns:

A Series containing the parameter space.

split_config() → Series[float]

Get the split configuration of the pipeline.

Returns:

A Series containing the split configuration.

train(G: Graph, **config: Any) → tuple[MODEL_TYPE, Series[Any]]

Train a model on a given graph using the pipeline.

Parameters:

• G – The graph to train on.

• **config – The configuration for training.

Returns:

A tuple containing the trained model and the result of the query.

train_estimate(G: Graph, **config: Any) → Series[Any]

Estimate the training time for a given graph and configuration.

Parameters:

• G – The graph to train on.

• **config – The configuration for training.

Returns:

The result of the query.

type() → str

Get the type of the pipeline.

Returns:

The type of the pipeline. It will be one of NodeClassificationPipeline, LinkPredictionPipeline, or NodeRegressionPipeline.