Hi, I received the following error when training the link prediction pipeline:
Failed to invoke procedure gds.beta.pipeline.linkPrediction.train: Caused by: java.lang.IllegalArgumentException: Node properties [xxx1, xxx2] defined in the feature steps do not exist in the graph or part of the pipeline.
I ran the gds.pipeline.list and got the following- there are node properties in the feature steps, but the nodePropertySteps is empty. Does it mean I need to add node properties into nodePropertySteps? If so, what is the syntax to add existing node property into nodepropertysteps please?
{
"splitConfig": {
"testFraction": 0.2,
"validationFolds": 3,
"trainFraction": 0.4,
"negativeSamplingRatio": 2.0
},
"autoTuningConfig": {
"maxTrials": 10
},
"featurePipeline": {
"nodePropertySteps": ,
"featureSteps": [
{
"name": "HADAMARD",
"config": {
"nodeProperties": [
"embedding",
"ProposalEncoding"
]
}
}
]
},
"trainingParameterSpace": {
"MultilayerPerceptron": ,
"RandomForest": ,
"LogisticRegression": [
{
"minEpochs": 1,
"maxEpochs": 5000,
"focusWeight": 0.0,
"patience": 2,
"tolerance": 0.001,
"learningRate": 0.001,
"batchSize": 100,
"penalty": {
"range": [
0.0001,
10000.0
]
},
"methodName": "LogisticRegression",
"classWeights":
}
]
}
}