k-Nearest Neighbour Classification of Datasets with a Family of Distances