Algorithms Overview

HiClass provides implementations for the most popular machine learning models for hierarchical classification, including the Local Classifier Per Node, the Local Classifier Per Parent Node and the Local Classifier Per Level. Additionally, the library includes metrics to evaluate model performance on hierarchical data. In this section we present in details the different approaches for hierarchical classification as well as the hierarchical metrics.

---

• Local Classifier Per Node
  • Training Policies
• Local Classifier Per Parent Node
• Local Classifier Per Level
• Multi-Label Classification
  • Motivation
  • Background - Classification Terminology
  • Design - Target Format
  • Fitting the Classifiers
    • Local Classifier Per Node
    • Local Classifier Per Parent Node
  • Prediction
    • Threshold
    • Tolerance
  • Metrics
  • Code example - Putting it all together
• Metrics
• Hierarchical Explainability
  • Motivation
  • Dataset overview
  • Background
  • API Design
  • Code sample
  • Filtering and Manipulation