HappieClust is an approximate version of agglomerative hierarchical clustering. When performing the standard full agglomerative hierarchical clustering, each pair of objects must be inspected to evaluate similarity. This is very time-consuming for large numbers of objects and/or complicated similarity measures. HappieClust performs agglomerative hierarchical clustering with partial information, not requiring all pairwise similarities to be known. HappieClust is further able to use similarity heuristics to carefully choose a subset of pairs for which the similarities are evaluated.