Andreas Zeitler03/30/2023, 11:47 AM
The output is configured in the data catalog. hook after_node_run_run:
node( modeloutput.predict, inputs=["estimator", "modelinput_x_" + name], outputs="modeloutput_" + name, tags=["output", "output_" + name] + tags, name="predict_" + name )
In the logs, it seems like kedro tries to load the data in the hook, before is was written by the catalog. Is this possible and is it meant to act like that? Could be fixed by using "outputs['modeloutput_dach_testsplit_test'] instead of catalog.load., but in my understanding it should not be necessary. Thanks in advance!
if node.name == 'predict_dach_testsplit_test': //This is the output of the node: y_pred_prob_comb_test = catalog.load('modeloutput_dach_testsplit_test') [..]
datajoely03/30/2023, 5:10 PM
Andreas Zeitler03/31/2023, 6:46 AM