Algorithm · Citadel · Hard
You have a labeled dataset for binary classification in which every example is an unordered collection of points rather than a single feature vector. Each point carries six numeric features. An example may contain anywhere from 16 to 100 points, and individual feature values can be absent, written as nan. Your job is to fit a model on the training examples and then predict a label for each test example. Whatever model you build has to meet four requirements: It accepts point…