Intrinsic dimensionality as a model-free measure of class imbalance

Key Points

• Intrinsic dimensionality serves as a model-free measure of class imbalance in datasets.
• Key evidence shows that varied intrinsic dimensionality affects algorithm performance metrics significantly.
• Assessment of different datasets demonstrates how intrinsic dimensionality relates to class distribution.
• Highlights the need for better understanding of data distribution in machine learning applications.