Pipelines

Forecast Pipelines

When launching a Forecast Pipeline within a recipe, specific details need to be provided to ensure the correct execution of the predictive model:

• pipeline_name: Identifies the particular pipeline configuration to be used. Mandatory.

• train_set: Specifies the dataset for model training. Mandatory.

• predict_set: Indicates the dataset on which predictions will be made. Mandatory.

Example

launch_forecast_pipeline:
  type: forecast_pipeline
  params:
    pipeline_name: sales_forecast
    train_set: train_dataset
    predict_set: test_dataset

Optimization Pipelines

Initiating an Optimization Pipeline requires identifying the dataset that comprises the optimization scenarios:

• pipeline_name: The unique identifier for the optimization configuration. Mandatory.

• optimization_set: The dataset containing the various options or scenarios for optimization. Mandatory.

Example

launch_optimization_pipeline:
  type: optimization_pipeline
  params:
    pipeline_name: Fleet_Optimization
    optimization_set: fleet_planning_system

Extract Pipeline Results

Different steps within the pipelines can be extracted as datasets using the import_from_pipeline method, providing flexibility in the output that you wish to analyze further. Learn more on extracting pipeline results.