job.yml

$schema: https://azuremlschemas.azureedge.net/latest/pipelineJob.schema.json
type: pipeline

# <inputs_and_outputs>
inputs:
  pipeline_job_input: #using dataset, can use datastore + datapath also
    dataset: 
      local_path: ./data

outputs: 
  pipeline_job_prepped_data:
    mode: upload
  pipeline_job_transformed_data:
    mode: upload   
  pipeline_job_trained_model:
    mode: upload
  pipeline_job_test_data:
    mode: upload 
  pipeline_job_predictions:
    mode: upload 
  pipeline_job_score_report:
    mode: upload
# </inputs_and_outputs>

# <jobs>
compute: azureml:cpu-cluster
settings:
  datastore: azureml:workspaceblobstore

jobs:
  prep-job:
    type: component
    component: file:./prep.yml
    inputs:
      raw_data: ${{inputs.pipeline_job_input}}
    outputs:
      prep_data: ${{outputs.pipeline_job_prepped_data}}

  transform-job:
    type: component
    component: file:./transform.yml
    inputs:
      clean_data: ${{jobs.prep-job.outputs.prep_data}}
    outputs:
      transformed_data: ${{outputs.pipeline_job_transformed_data}}

  train-job:
    type: component
    component: file:./train.yml
    inputs:
      training_data: ${{jobs.transform-job.outputs.transformed_data}}
    outputs:
      model_output: ${{outputs.pipeline_job_trained_model}}
      test_data: ${{outputs.pipeline_job_test_data}}
  
  predict-job:
    type: component
    component: file:./predict.yml
    inputs:
      model_input: ${{jobs.train-job.outputs.model_output}}
      test_data: ${{jobs.train-job.outputs.test_data}}
    outputs:
      predictions: ${{outputs.pipeline_job_predictions}}

  score-job:
    type: component
    component: file:./score.yml
    inputs:
      predictions: ${{jobs.predict-job.outputs.predictions}}
      model: ${{jobs.train-job.outputs.model_output}}
    outputs:
      score_report: ${{outputs.pipeline_job_score_report}}
# </jobs>