Clustering models in machine learning must be assessed by how well they separate data into meaningful groups with distinctive characteristics.