Warning

This documentation is for an old version (v2.5.0) of cleanlab. To see the documentation for the latest stable version (v2.7.1), click here.

Datalab guides

Guides for using Datalab and understanding the issues it detects.

Note

Using Datalab requires additional dependencies beyond the rest of the cleanlab package. To install them, run:

$ pip install "cleanlab[datalab]"

For the developmental version of the package, install from source:

$ pip install "git+https://github.com/cleanlab/cleanlab.git#egg=cleanlab[datalab]"

Types of issues

Guides to use Datalab with greater control, selecting what issues to search for and what nondefault settings to use for detecting them.

• Datalab Issue Types
  • Types of issues Datalab can detect
    • Estimates for Each Issue Type
    • Label Issue
    • Outlier Issue
    • (Near) Duplicate Issue
    • Non-IID Issue
    • Class-Imbalance Issue
    • Image-specific Issues
  • Optional Issue Parameters
    • Label Issue Parameters
    • Outlier Issue Parameters
    • Duplicate Issue Parameters
    • Non-IID Issue Parameters
    • Imbalance Issue Parameters

Customizing issue types

Guides (for developers) to create a custom issue type that Datalab audits for together with its built-in issue types.

• Creating Your Own Issues Manager
  • Prerequisites
  • Implementing IssueManagers
    • Basic Issue Check
    • Intermediate Issue Check
    • Advanced Issue Check
  • Use with Datalab