implicit_als_matrix_factorization_user_item_recommendations.py

1import implicit
2from scipy.sparse import csr_matrix
3import numpy as np
4
5# Create a random sparse matrix of user/item confidence weights
6# In a real app, this would be your user-item interaction data
7data = np.array([1, 2, 3, 4, 5, 6])
8rows = np.array([0, 0, 1, 2, 2, 2])
9cols = np.array([0, 2, 1, 0, 1, 2])
10user_items = csr_matrix((data, (rows, cols)), shape=(3, 3))
11
12# Initialize an Alternating Least Squares model
13model = implicit.als.AlternatingLeastSquares(factors=64, iterations=10)
14
15# Train the model on the user-item matrix
16model.fit(user_items)
17
18# Recommend items for a user (userid 0)
19# Returns a list of (itemid, score) tuples
20userid = 0
21recommendations = model.recommend(userid, user_items[userid])
22
23print(f"Recommendations for user {userid}: {recommendations}")