Causal Discovery from Subsampled Time Series Data by Constraint Optimization.

Area of Science:

• Causal inference
• Time series analysis
• Systems biology

Background:

• Subsampling time series data can distort causal structure estimation.
• Accurate causal discovery is crucial for understanding complex systems.

Purpose of the Study:

• To develop methods for estimating causal structures from time series data sampled at a coarser timescale than the underlying system dynamics.
• To improve the accuracy and robustness of causal inference in the presence of subsampling.

Main Methods:

• Developed a constraint satisfaction procedure for mapping measurement timescale structures to system timescale causal structures.
• Proposed a novel constraint optimization algorithm for recovering system timescale causal structures from finite-sample data.
• The new methods offer significant computational improvements and optimal recovery from statistical errors.

Main Results:

• The constraint satisfaction procedure shows orders of magnitude better performance than prior methods.
• The constraint optimization approach is the first to optimally recover system timescale causal structures from subsampled data.
• The proposed algorithms provide robust and non-parametric causal structure estimation.

Conclusions:

• Accurate causal structure estimation from subsampled time series is achievable with advanced methods.
• The developed algorithms enhance the reliability of causal inference in undersampled systems.
• This work offers a significant advancement for analyzing complex systems with limited temporal resolution.