I have a KNN, and I need to partially argsort a list. Here is how it works right now in code: sorted_distance_indices = distances.argsort(axis=1)[:,:self.parameters['k']+1] kplus_1_nearest_classes = self.trainingY[sorted_distance_indices] ...etc. I found this answer, How can I partially sort a list? But I don't see how to adapt the 'heapification' for the argsort task, (and I have no idea how to do language interop in Python, so I don't see how to do the heapsort alg manually)... Continue reading...
