TemporalSplitPreprocessing

Temporal split preprocessing step with customer-level splitting and OOT validation

SageMaker step type	Processing
Node type	internal (consumes upstream, produces downstream)
Container entry point	temporal_split_preprocessing.py
Interface file	steps/interfaces/temporal_split_preprocessing.step.yaml

Compute

Compute kind	sklearn

Functionality

Temporal split preprocessing script. Handles data loading, temporal splitting, customer-level splitting, and main task label generation.

Inputs (dependencies)

Input	Type	Required	Compatible producers
DATA	processing_output	yes	CradleDataLoading, DummyDataLoading, DataLoad, ProcessingStep
SIGNATURE	processing_output	no	CradleDataLoading, DummyDataLoading

Outputs

Output	Type
training_data	training_data
oot_data	processing_output

Consumers (downstream steps)

Steps that declare this step as a compatible input source:

• LightGBMMTModelEval

• LightGBMMTTraining

• XgboostMtModelEval

• XgboostMtTraining

Framework requirements

Package	Version
pandas	>=1.3.0
numpy	>=1.21.0

---

← Back to the Step Catalog