Data Generating Process to Evaluate Causal Discovery Techniques for Time Series Data

causaLens’ NeurIPS 2020 paper sets out a framework for benchmarking causal discovery techniques time series data.

causaLens researchers Andrew Lawrence, Marcus Kaiser, Rui Sampaio, and Maksim Sipos introduce a novel framework for evaluating and benchmarking causal discovery methods for time-series data. The paper — which also evaluates prominent causal discovery algorithms, and sets out how the framework can support researchers and data science practitioners — was presented at leading AI conference NeurIPS.