Supremum Metric in Python for Knn with Uncertain Data

Question

I'm trying to make a classifier for uncertain data (e.g ranged data) using python. in certain dataset, the list is a 2D array or array of record (contains float numbers for data and a string for labels), where in uncertain dataset the list is a 3D array (contains range of float numbers for data and a string for labels). i managed to manipulate a certain dataset to be uncertain using uniform probability distribution. A research paper says that i have to use supremum distance metric. how do i implement this metric in python? note that in uncertain dataset, both test set and training set is uncertain

Answer 1

I found out using scipy spatial distance and tweaking for-loops in standard knn helps a lot