In clustering, providing an explanation of the results is an important task. Pattern-based clustering algorithms provide, in addition to the list of objects belonging to each cluster, an explanation of the results in terms of a set of patterns that describe the objects grouped in each cluster. It makes these algorithms very attractive from the practical point of view; however, patternbased clustering algorithms commonly have a high computational cost in the clustering stage. Moreover, the most recent algorithms proposed within this approach, extract patterns from numerical datasets by applying an a priori discretization process, which may cause information loss. In this thesis, we propose new algorithms for extracting only a subset of patterns useful for clustering, from a collection of diverse unsupervised decision trees induced from a dataset. Additionally, we propose a new clustering algorithm based on these patterns.