Surprise

Surprise is a Python scikit designed for building and analyzing recommender systems based on explicit rating data. It offers a variety of ready-to-use prediction algorithms (including SVD, NMF, k-NN, and others), various similarity measures, and robust tools for dataset handling, performance evaluation, and cross-validation. The library targets researchers, data scientists, and developers who wish to experiment with, build, and benchmark collaborative filtering models for recommendation tasks.