search

microsoft / solution-accelerator-many-models

MIT License
193 stars 85 forks source link

copy_predictions.py fails because of pandas? #141

Closed jknmsft closed 2 years ago

jknmsft commented 2 years ago

I am trying to run the pipeline in 03_CustomScript_Forecasting_Pipeline.ipynb but if fails when it gets to import pandas as pd in the copy_predictions.py script. Seems strange since the forecast.py notebook that precedes it in the pipeline also has an import pandas as pd statement and it works

The CondaDependencies.create arguments has pandas in it, see below

forecast_conda_deps = CondaDependencies.create(pip_packages=['sklearn', 'pandas', 'joblib', 'azureml-defaults', 'azureml-core', 'azureml-dataprep[fuse]'])

jknmsft commented 2 years ago

Solved it myself, added a runconfig in the arguments for PythonScriptStep