Dependency failure running 03_CustomScript Forecast

[parasin]

Steup and all previous notebooks run ok. Then at the Copy Predictions PythonScript step you get a failure with the following standard error: "[stderr]Traceback (most recent call last):
[stderr] File "copy_predictions.py", line 4, in
[stderr] import pandas as pd
[stderr]ModuleNotFoundError: No module named 'pandas'
[stderr]

Environment or Runconfig isn't set for this.

With some investigation error is at the Copy Predictions step in the 03_CustomScriptForecast notebook. I was able to remediate by creating a RunConfig and adding that to the PythonScriptStep.

from azureml.core.conda_dependencies import CondaDependencies
from azureml.core import RunConfiguration

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

The following is the same but now with the added runconfig argument being set to the object

upload_predictions_step = PythonScriptStep(
name="copy_predictions",
script_name="copy_predictions.py",
compute_target=compute,
source_directory='./scripts',
inputs=[output_dref, output_dir],
runconfig=run_config,
allow_reuse=False,
arguments=['--parallel_run_step_output', output_dir,
'--output_dir', output_dref,
'--target_column', 'Quantity',
'--timestamp_column', 'WeekStarting',
'--timeseries_id_columns', 'Store', 'Brand']