Machine Learning Algorithms - Second Edition

Implementing a user-based system with scikit-learn

For our purposes, we need to create a dummy dataset of users and products (but it's very easy to employ real datasets):

import numpy as npnb_users = 1000users = np.zeros(shape=(nb_users, 4))for i in range(nb_users):   users[i, 0] = np.random.randint(0, 4)    users[i, 1] = np.random.randint(0, 2)    users[i, 2] = np.random.randint(0, 5)    users[i, 3] = np.random.randint(0, 5)

We assume that we have 1000 users with 4 features, represented by integer numbers bounded between 0 and 4 or 5. It doesn't matter what they mean; their role is to characterize a user and to allow for the clustering of the set. Of course, the features must be consistent with the task. In other words, we are assuming that the distances ...