Causality-Guided Feature Selection

Extreme Mei-yu rainfall (MYR) can cause catastrophic impacts to the economic development and societal welfare in China. While significant improvements have been made in climate models, they often struggle to simulate local-to-regional extreme rainfall (e.g., MYR). Yet, large-scale climate modes (LSCMs) are relatively well represented in climate models. Since there exists a close relationship between MYR and various LSCMs, it might be possible to develop causality-guided statistical models for MYR prediction based on LSCMs. These statistical models could then be applied to climate model simulations to improve the representation of MYR in climate models.In this pilot study, it is demonstrated that skillful causality-guided statistical models for MYR can be constructed based on known LSCMs. The relevancy of the selected predictors for statistical models are found to be consistent with the literature. The importance of temporal resolution in constructing statistical models for MYR is also shown and is in good agreement with the literature. The results demonstrate the reliability of the causality-guided approach in studying complex circulation systems such as the East Asian summer monsoon (EASM). Some limitations and possible improvements of the current approach are discussed. The application of the causality-guided approach opens up a new possibility to uncover the complex interactions in the EASM in future studies.

show abstract

Causality-Guided Feature Selection

Cited by 4 publications

References 13 publications

Recognizing Non-small Cell Lung Cancer Subtypes by a Constraint-Based Causal Network from CT Images

Recognizing Non-small Cell Lung Cancer Subtypes by a Constraint-Based Causal Network from CT Images

A Community-Driven Graph Partitioning Method for Constraint-Based Causal Discovery

A Causality-guided Statistical Approach for Modeling Extreme Mei-yu Rainfall Based on Known Large-scale Modes—A Pilot Study

Contact Info

Product

Resources

About