Statistically validated hierarchical clustering: Nested partitions in hierarchical trees

We develop a greedy algorithm that is fast and scalable in the detection of a nested partition extracted from a dendrogram obtained from hierarchical clustering of a multivariate series. Our algorithm provides a p-value for each clade observed in the hierarchical tree. The p-value is obtained by computing a number of bootstrap replicas of the dissimilarity matrix and by performing a statistical test on each difference between the dissimilarity associated with a given clade and the dissimilarity of the clade of its parent node. We prove the efficacy of our algorithm with a set of benchmarks generated by using a hierarchical factor model. We compare the results obtained by our algorithm with those of Pvclust. Pvclust is a widely used algorithm developed with a global approach originally motivated by phylogenetic studies. In our numerical experiments we focus on the role of multiple hypothesis test correction and on the robustness of the algorithms to inaccuracy and errors of datasets. We also apply our algorithm to a reference empirical dataset. We verify that our algorithm is much faster than Pvclust algorithm and has a better scalability both in the number of elements and in the number of records of the investigated multivariate set. Our algorithm provides a hierarchically nested partition in much shorter time than currently widely used algorithms allowing to perform a statistically validated cluster analysis detection in very large systems.