DenseBlocks is an unsupervised classification method for n-dimensional numerical data. It creates a multi-dimensional density representation of given training samples, defines clusters accordingly and returns this representation which can further be used to classify related data.