opal is an osu!mania player accuracy-prediction model.
Opal is a Machine Learning model that uses a Matrix Factorization branch then a Multi-layered Perceptron branch to learn associations between user and maps. Then use those learnt associations to predict new scores never before seen. It uses the idea from NeuMF, also known as Neural Collaborative Filtering
He, X., Liao, L., Zhang, H., Nie, L., Hu, X., & Chua, T. S. (2017, April). Neural collaborative filtering. In Proceedings of the 26th international conference on world wide web (pp. 173-182).
It’s also dependent on osu-data which is a bash script wrapped in a Python Package. It uses Docker Containers to automatically downloads the source database data, optimally loads it into a MySQL container.
Read more about osu-data in our own article here