Hierarchical Variable Clustering Based on Measures of Predictability

Within a portfolio of random variables, complex structures and associations can occur that influence the overall behaviour of the portfolio. For revealing and identifying such interrelationships, a variable cluster analysis can be performed as a pre-processing step in a data analysis. Motivated by recent developments in the field of hierarchical variable clustering, we here present a short and concise summary of the article Hierarchical variable clustering based on the predictive strength between random vectors [5], in which a variable clustering procedure is introduced that is based on a measure of predictability.