Publisher Correction: Causal inference for time series

Jakob Runge, Andreas Gerhardus, Gherardo Varando, Veronika Eyring, Gustau Camps‐VallsNature Reviews Earth & Environment 2023 · DOI

TL;DR

This publisher correction addresses errors in a previously published article on causal inference for time series. The correction clarifies that a crucial red trace representing MLP-causal was missing from a key figure and that a symbol indicating dependence was incorrectly displayed. These corrections are vital for accurately interpreting the article's findings on causal relationships within time-series data.

What Problem Does This Paper Address?

This document is a publisher's correction to a prior article titled "Causal inference for time series." The original article, published in Nature Reviews Earth & Environment, aimed to discuss methods for inferring causal relationships from time-series data. Causal inference in time series is a challenging problem because correlation does not imply causation, and temporal dependencies can be complex. Distinguishing true causal links from mere correlations is crucial for understanding dynamic systems, predicting future behavior, and designing effective interventions in fields ranging from climate science to economics and neuroscience. The correction specifically addresses inaccuracies in the presentation of results related to a method called "MLP-causal" and the interpretation of symbols used to denote statistical dependence.

Methods

The correction itself does not introduce new methods. Instead, it rectifies errors in the presentation of methods and results from the original article. The original article likely employed or discussed various causal inference techniques applicable to time series, such as Granger causality, transfer entropy, or machine learning-based approaches like the MLP-causal method mentioned. The correction highlights that the MLP-causal method's results were misrepresented in a figure, and the interpretation of a symbol indicating dependence was also flawed.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Importance of Accurate Visual Representation: The correction underscores how critical it is for figures to accurately represent the data and results. The omission of the "red trace for MLP-causal" in Figure 4i likely obscured a significant part of the analysis or comparison.
  • Precision in Notation: The misrepresentation of the symbol for "dependent with" highlights the need for meticulous attention to detail in scientific notation, as even small errors can lead to misinterpretations of relationships between variables.
  • Reproducibility and Verifiability: Such corrections are essential for ensuring the reproducibility and verifiability of scientific findings. Readers relying on the original publication might have drawn incorrect conclusions without these clarifications.

Key Results

The correction states that:

  • The red trace for MLP-causal was missing from Figure 4i in the initially published version of the article.
  • In the text following equation (2), the symbol for "dependent with" was incorrectly displayed, and it should have indicated independence.

These errors have since been corrected in the HTML and PDF versions of the article.

Limitations and Future Work

Not discussed in the paper.

Summary updated

BibTeX
@article{runge2023publisher,
  author = {Jakob Runge and Andreas Gerhardus and Gherardo Varando and Veronika Eyring and Gustau Camps‐Valls},
  title = {{Publisher Correction: Causal inference for time series}},
  journal = {Nature Reviews Earth \& Environment},
  year = {2023},
  doi = {10.1038/s43017-023-00471-4},
}
Figure