Korrelaten

Korrelaten is a term used in statistics and data analysis to describe variables that show a statistical association with a specified target variable. The concept is closely related to, but distinct from, correlation in the sense of a single coefficient; korrelaten emphasizes the collection of multiple variables that move with the target variable, potentially in both positive and negative directions. The term is derived from the Dutch word cor-relatie and is used mainly in Dutch-language statistical literature.

Korrelaten are identified during exploratory data analysis or feature selection. Common methods include calculating correlation coefficients (such as Pearson's r or Spearman's rho) between each candidate variable and the target; for more complex relationships, information-theoretic measures like mutual information can be used. In multivariate contexts, partial correlations are used to control for the influence of other variables, revealing direct associations.

Interpretation of korrelaten requires caution: correlation does not imply causation, and spurious correlations can arise from confounding, selection bias, or multiple testing. In high-dimensional data, strategies such as false discovery rate control, cross-validation, and regularization help mitigate overfitting when identifying korrelaten.

Applications span social sciences, epidemiology, marketing, and finance, where researchers seek to understand factors that accompany or predict outcomes. The term remains informal in some regions, with correlates or associated variables commonly used in English-language literature. Nevertheless, korrelaten serves as a practical label for the set of variables that cluster around a target in statistical analyses.