Topological Drift Detection Benchmark with Seven Regime Types

This dataset serves as a benchmark for drift detection using topological data analysis. It contains seven distinct regime types, some of which are designed to be undetectable by traditional statistical methods but detectable via persistent homology.

Use Cases

• Benchmark persistent homology methods against seven defined regime types for drift detection in time series.
• Compare the sensitivity of topological and statistical methods to regime changes, focusing on regimes invisible to statistics.
• Develop and validate new topological data analysis algorithms for identifying specific regime types in sequential data.

Strengths

• Includes seven distinct regime types for systematic evaluation.
• Focuses on regimes specifically designed to challenge traditional statistical detection methods.

Limitations

• The dataset size, number of rows, and specific features are unknown, limiting assessment of scale and applicability.
• The absence of column details prevents understanding of the data structure and feature availability.