HAIPipe ML Data Preparation

As Layer 4 of the HAIPipe research ecosystem, this skill bridges the gap between raw feature extraction and model training by converting CaseSets into standardized AIDataSets. It provides a robust framework for handling train/validation/test splitting, temporal encoding (TEToken), and complex input/output transforms while maintaining metadata lineage. Built to be domain-agnostic, it allows researchers to systematically prepare data for domains ranging from healthcare (EHR/CGM) to NLP using a configuration-driven 'cooking' metaphor.

주요 기능

사용 사례