Explanation¶

Algorithm Summary¶

For a quick reference of each algorithm’s functionality and requirements, see the Algorithm Summary page.

Concepts¶

These explanations dive into the what, why and theory behind each of DataEval’s features.

• Balance

• Bayes Error Rate

• Clustering

• Coverage

• Completeness

• Dataset Cleaning

• Dataset Prioritization

• Divergence

• Diversity

• Drift

• Embeddings

• Image Statistical Analysis

• Label Parity

• Metadata

• Model Training

• Out-of-Distribution Data

• Outliers

• Parity

• Sufficiency of Data

• Upperbound on Average Precision

Workflows¶

The Operational Machine Learning Lifecycle has multiple stages that are interwoven to create robust, reliable, and generalizable models. Here you will learn about each stage, how they interact, and how they differ from competition-based machine learning life cycles.

The workflows described in this document give an overview to the role-specific stages that you might be tasked with. It is important to note that these workflows are flexible, and many times multiple hats are needed in a professional setting.

We have created a few entry-level guides for common roles found in Data Analytics and machine learning to develop the crucial knowledge one must know to become an expert practitioner.

• Testing & Evaluation Engineer

• Artificial Intelligence Engineer