Abstract: Highlights•Our proposed feature selection method is classified as unsupervised, filter and multivariate.•The possible dependencies between features are considered to reduce the redundancy among the selected features.•The proposed method manages the trade-off between computational time and quality of the results.•The method has been compared to well-known univariate and multivariate methods on the different classifiers.•The experimental results indicate that the method outperforms the unsupervised methods and is comparable with the supervised methods.