def test_transforms_twoheterogeneousmodelcombiner(self):
# import pdb; pdb.set_trace()
# args
transform_model = "$transform_model"
predictor_model = "$predictor_model"
model = "$output_model"
# call
node = transforms_twoheterogeneousmodelcombiner(
transform_model=transform_model, predictor_model=predictor_model,
model=model)
# check
assert isinstance(node, EntryPoint)
assert node.inputs["TransformModel"] == transform_model
assert node.inputs["PredictorModel"] == predictor_model
assert node.outputs["PredictorModel"] == model
assert node.input_variables == {transform_model, predictor_model}
assert node.output_variables == {model}
def test_logistic_regression_graph(self):
# import pdb; pdb.set_trace()
# args
data = "$input_data"
features = ["xint1"]
output_data = "$training_data"
model = "$transform_model"
# call
feature_node = transforms_featurecombiner(data=data,
features=features,
output_data=output_data,
model=model)
# args
training_data = "$training_data"
quiet = False
label_column = "ylogical"
predictor_model = "$predictor_model"
# call
lr_node = trainers_logisticregressionbinaryclassifier(
# , FeatureColumn = "Features"
training_data=training_data,
quiet=quiet,
label_column=label_column,
predictor_model=predictor_model)
# args
transform_model = "$transform_model"
predictor_model = "$predictor_model"
model = "$output_model"
# call
combine_node = transforms_twoheterogeneousmodelcombiner(
transform_model=transform_model,
predictor_model=predictor_model,
model=model)
# compose graph
# graph_sub = Graph(feature_node, lr_node, combine_node)
# print(graph_sub)
all_nodes = [feature_node, lr_node, combine_node]
graph = Graph(dict(input_data=""), dict(output_model=""), False,
*all_nodes)
# print(graph)
graph.run(X=None, dryrun=True)