Where to Go Next¶
Choose your path¶
Not sure where to begin? Pick the description that fits your situation:
New to DataEval
Never used DataEval before. The Beginner’s Guide walks through the basics of downloading DataEval, setting up your dataset and checking your dataset for duplicates and outliers — no prior DataEval experience needed.
Know what problem you have
Use the Which Tool Guide to find the right evaluator for your situation, organized by the question you’re trying to answer.
Know what tool you need
Go straight to the Functional Overview for a quick-reference table of each algorithm’s inputs, outputs, and task applicability.
Learn by doing¶
Tutorials
End-to-end walkthroughs for common T&E tasks. Start here if you want guided practice — each tutorial takes you from a dataset to a decision.
How-To Guides
Focused, goal-oriented guides for specific tasks. Start here if you know what you need to do and want the steps without the explanation.
Understand the concepts¶
Concept Pages
In-depth explanations of the theory behind each evaluator — what it measures, why it matters, and how to interpret what it finds.
Roles Guide
Role-specific guidance for data scientists, ML engineers, and T&E engineers, organized around the analysis tasks you encounter day-to-day.
Look things up¶
API Reference
Complete technical documentation for every function and class in DataEval.
Glossary
Definitions of key terms used throughout the documentation.
JATIC Interoperability¶
DataEval is part of the Joint AI Test Infrastructure Capability (JATIC) product suite for AI Test & Evaluation. The goal of the program is to deliver a suite of tools that interoperate using a standardized interface that enables modular, reproducible assessments for ease-of-use, performance and reliability.
MAITE (Modular AI Trustworthy Engineering): Standardized protocols for AI T&E workflows. https://mit-ll-ai-technology.github.io/maite/
NRTK (Natural Robustness Toolkit): Perturbation toolkit for applying realistic perturbations to evaluate model natural robustness. https://nrtk.readthedocs.io/
XAITK-Saliency (Explainable AI Toolkit): Explainability toolkit for understanding model decisions. https://xaitk-saliency.readthedocs.io/