[![GitHub version](https://badge.fury.io/gh/nicolashug%2FSurprise.svg)](https://badge.fury.io/gh/nicolashug%2FSurprise) [![Documentation Status](https://readthedocs.org/projects/surprise/badge/?version=stable)](https://surprise.readthedocs.io/en/stable/?badge=stable) [![python versions](https://img.shields.io/badge/python-3.8+-blue.svg)](https://surpriselib.com) [![License](https://img.shields.io/badge/License-BSD%203--Clause-blue.svg)](https://opensource.org/licenses/BSD-3-Clause) [![DOI](https://joss.theoj.org/papers/10.21105/joss.02174/status.svg)](https://doi.org/10.21105/joss.02174) [![logo](./logo_black.svg)](https://surpriselib.com) Overview -------- [Surprise](https://surpriselib.com) is a Python [scikit](https://projects.scipy.org/scikits.html) for building and analyzing recommender systems that deal with explicit rating data. [Surprise](https://surpriselib.com) **was designed with the following purposes in mind**: - Give users perfect control over their experiments. To this end, a strong emphasis is laid on [documentation](https://surprise.readthedocs.io/en/stable/index.html), which we have tried to make as clear and precise as possible by pointing out every detail of the algorithms. - Alleviate the pain of [Dataset handling](https://surprise.readthedocs.io/en/stable/getting_started.html#load-a-custom-dataset). Users can use both *built-in* datasets ([Movielens](https://grouplens.org/datasets/movielens/), [Jester](https://eigentaste.berkeley.edu/dataset/)), and their own *custom* datasets. - Provide various ready-to-use [prediction algorithms](https://surprise.readthedocs.io/en/stable/prediction_algorithms_package.html) such as [baseline algorithms](https://surprise.readthedocs.io/en/stable/basic_algorithms.html), [neighborhood methods](https://surprise.readthedocs.io/en/stable/knn_inspired.html), matrix factorization-based ( [SVD](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVD), [PMF](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#unbiased-note), [SVD++](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVDpp), [NMF](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.NMF)), and [many others](https://surprise.readthedocs.io/en/stable/prediction_algorithms_package.html). Also, various [similarity measures](https://surprise.readthedocs.io/en/stable/similarities.html) (cosine, MSD, pearson...) are built-in. - Make it easy to implement [new algorithm ideas](https://surprise.readthedocs.io/en/stable/building_custom_algo.html). - Provide tools to [evaluate](https://surprise.readthedocs.io/en/stable/model_selection.html), [analyse](https://nbviewer.jupyter.org/github/NicolasHug/Surprise/tree/master/examples/notebooks/KNNBasic_analysis.ipynb/) and [compare](https://nbviewer.jupyter.org/github/NicolasHug/Surprise/blob/master/examples/notebooks/Compare.ipynb) the algorithms' performance. Cross-validation procedures can be run very easily using powerful CV iterators (inspired by [scikit-learn](https://scikit-learn.org/) excellent tools), as well as [exhaustive search over a set of parameters](https://surprise.readthedocs.io/en/stable/getting_started.html#tune-algorithm-parameters-with-gridsearchcv). The name *SurPRISE* (roughly :) ) stands for *Simple Python RecommendatIon System Engine*. Please note that surprise does not support implicit ratings or content-based information. Getting started, example ------------------------ Here is a simple example showing how you can (down)load a dataset, split it for 5-fold cross-validation, and compute the MAE and RMSE of the [SVD](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVD) algorithm. ```python from surprise import SVD from surprise import Dataset from surprise.model_selection import cross_validate # Load the movielens-100k dataset (download it if needed). data = Dataset.load_builtin('ml-100k') # Use the famous SVD algorithm. algo = SVD() # Run 5-fold cross-validation and print results. cross_validate(algo, data, measures=['RMSE', 'MAE'], cv=5, verbose=True) ``` **Output**: ``` Evaluating RMSE, MAE of algorithm SVD on 5 split(s). Fold 1 Fold 2 Fold 3 Fold 4 Fold 5 Mean Std RMSE (testset) 0.9367 0.9355 0.9378 0.9377 0.9300 0.9355 0.0029 MAE (testset) 0.7387 0.7371 0.7393 0.7397 0.7325 0.7375 0.0026 Fit time 0.62 0.63 0.63 0.65 0.63 0.63 0.01 Test time 0.11 0.11 0.14 0.14 0.14 0.13 0.02 ``` [Surprise](https://surpriselib.com) can do **much** more (e.g, [GridSearchCV](https://surprise.readthedocs.io/en/stable/getting_started.html#tune-algorithm-parameters-with-gridsearchcv))! You'll find [more usage examples](https://surprise.readthedocs.io/en/stable/getting_started.html) in the [documentation ](https://surprise.readthedocs.io/en/stable/index.html). Benchmarks ---------- Here are the average RMSE, MAE and total execution time of various algorithms (with their default parameters) on a 5-fold cross-validation procedure. The datasets are the [Movielens](https://grouplens.org/datasets/movielens/) 100k and 1M datasets. The folds are the same for all the algorithms. All experiments are run on a laptop with an intel i5 11th Gen 2.60GHz. The code for generating these tables can be found in the [benchmark example](https://github.com/NicolasHug/Surprise/tree/master/examples/benchmark.py). | [Movielens 100k](http://grouplens.org/datasets/movielens/100k) | RMSE | MAE | Time | |:---------------------------------------------------------------------------------------------------------------------------------------|-------:|------:|:--------| | [SVD](http://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVD) | 0.934 | 0.737 | 0:00:06 | | [SVD++ (cache_ratings=False)](http://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVDpp) | 0.919 | 0.721 | 0:01:39 | | [SVD++ (cache_ratings=True)](http://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVDpp) | 0.919 | 0.721 | 0:01:22 | | [NMF](http://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.NMF) | 0.963 | 0.758 | 0:00:06 | | [Slope One](http://surprise.readthedocs.io/en/stable/slope_one.html#surprise.prediction_algorithms.slope_one.SlopeOne) | 0.946 | 0.743 | 0:00:09 | | [k-NN](http://surprise.readthedocs.io/en/stable/knn_inspired.html#surprise.prediction_algorithms.knns.KNNBasic) | 0.98 | 0.774 | 0:00:08 | | [Centered k-NN](http://surprise.readthedocs.io/en/stable/knn_inspired.html#surprise.prediction_algorithms.knns.KNNWithMeans) | 0.951 | 0.749 | 0:00:09 | | [k-NN Baseline](http://surprise.readthedocs.io/en/stable/knn_inspired.html#surprise.prediction_algorithms.knns.KNNBaseline) | 0.931 | 0.733 | 0:00:13 | | [Co-Clustering](http://surprise.readthedocs.io/en/stable/co_clustering.html#surprise.prediction_algorithms.co_clustering.CoClustering) | 0.963 | 0.753 | 0:00:06 | | [Baseline](http://surprise.readthedocs.io/en/stable/basic_algorithms.html#surprise.prediction_algorithms.baseline_only.BaselineOnly) | 0.944 | 0.748 | 0:00:02 | | [Random](http://surprise.readthedocs.io/en/stable/basic_algorithms.html#surprise.prediction_algorithms.random_pred.NormalPredictor) | 1.518 | 1.219 | 0:00:01 | | [Movielens 1M](https://grouplens.org/datasets/movielens/1m) | RMSE | MAE | Time | |:----------------------------------------------------------------------------------------------------------------------------------------|-------:|------:|:--------| | [SVD](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVD) | 0.873 | 0.686 | 0:01:07 | | [SVD++ (cache_ratings=False)](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVDpp) | 0.862 | 0.672 | 0:41:06 | | [SVD++ (cache_ratings=True)](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVDpp) | 0.862 | 0.672 | 0:34:55 | | [NMF](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.NMF) | 0.916 | 0.723 | 0:01:39 | | [Slope One](http://surprise.readthedocs.io/en/stable/slope_one.html#surprise.prediction_algorithms.slope_one.SlopeOne) | 0.907 | 0.715 | 0:02:31 | | [k-NN](http://surprise.readthedocs.io/en/stable/knn_inspired.html#surprise.prediction_algorithms.knns.KNNBasic) | 0.923 | 0.727 | 0:05:27 | | [Centered k-NN](http://surprise.readthedocs.io/en/stable/knn_inspired.html#surprise.prediction_algorithms.knns.KNNWithMeans) | 0.929 | 0.738 | 0:05:43 | | [k-NN Baseline](http://surprise.readthedocs.io/en/stable/knn_inspired.html#surprise.prediction_algorithms.knns.KNNBaseline) | 0.895 | 0.706 | 0:05:55 | | [Co-Clustering](http://surprise.readthedocs.io/en/stable/co_clustering.html#surprise.prediction_algorithms.co_clustering.CoClustering) | 0.915 | 0.717 | 0:00:31 | | [Baseline](http://surprise.readthedocs.io/en/stable/basic_algorithms.html#surprise.prediction_algorithms.baseline_only.BaselineOnly) | 0.909 | 0.719 | 0:00:19 | | [Random](http://surprise.readthedocs.io/en/stable/basic_algorithms.html#surprise.prediction_algorithms.random_pred.NormalPredictor) | 1.504 | 1.206 | 0:00:19 | Installation ------------ With pip: $ pip install scikit-surprise With conda: $ conda install -c conda-forge scikit-surprise For the latest version, you can also clone the repo and build the source (you'll first need [Cython](https://cython.org/) and [numpy](https://www.numpy.org/)): $ git clone https://github.com/NicolasHug/surprise.git $ cd surprise $ pip install . License and reference --------------------- This project is licensed under the [BSD 3-Clause](https://opensource.org/licenses/BSD-3-Clause) license, so it can be used for pretty much everything, including commercial applications. I'd love to know how Surprise is useful to you. Please don't hesitate to open an issue and describe how you use it! Please make sure to cite the [paper](https://joss.theoj.org/papers/10.21105/joss.02174) if you use Surprise for your research: @article{Hug2020, doi = {10.21105/joss.02174}, url = {https://doi.org/10.21105/joss.02174}, year = {2020}, publisher = {The Open Journal}, volume = {5}, number = {52}, pages = {2174}, author = {Nicolas Hug}, title = {Surprise: A Python library for recommender systems}, journal = {Journal of Open Source Software} } Contributors ------------ The following persons have contributed to [Surprise](https://surpriselib.com): ashtou, Abhishek Bhatia, bobbyinfj, caoyi, Chieh-Han Chen, Raphael-Dayan, Олег Демиденко, Charles-Emmanuel Dias, dmamylin, Lauriane Ducasse, Marc Feger, franckjay, Lukas Galke, Tim Gates, Pierre-François Gimenez, Zachary Glassman, Jeff Hale, Nicolas Hug, Janniks, jyesawtellrickson, Doruk Kilitcioglu, Ravi Raju Krishna, lapidshay, Hengji Liu, Ravi Makhija, Maher Malaeb, Manoj K, James McNeilis, Naturale0, nju-luke, Pierre-Louis Pécheux, Jay Qi, Lucas Rebscher, Craig Rodrigues, Skywhat, Hercules Smith, David Stevens, Vesna Tanko, TrWestdoor, Victor Wang, Mike Lee Williams, Jay Wong, Chenchen Xu, YaoZh1918. Thanks a lot :) ! Development Status ------------------ Starting from version 1.1.0 (September 2019), I will only maintain the package, provide bugfixes, and perhaps sometimes perf improvements. I have less time to dedicate to it now, so I'm unabe to consider new features. For bugs, issues or questions about [Surprise](https://surpriselib.com), please avoid sending me emails; I will most likely not be able to answer). Please use the GitHub [project page](https://github.com/NicolasHug/Surprise) instead, so that others can also benefit from it.