Open source machine learning in computational chemistry

For each project, Hagg and Kirschner provide a short description, a link to the code, the associated licence type and whether the training data and the resulting models are publicly available. Based on the projects deposited GitHub repositories, the most frequently used Python libraries are indentified. This overview will serve as a resource to learn more about how machine learning can be applied to computational chemistry topics by identifying accessible codes with accompanying papers on a topic-by-topic basis. To this end, open source computational chemistry software for generating training data and basic Python machine learning libraries were also included. Based on the observations and considering the three pillars of collaborative work in machine learning - open data, open sources  code and open models - the researchers make some suggestions to the community.