All available crystallographic information is used to train classification models for crystal systems, Bravais lattice, point groups and space groups. Reasonably accurate results can be achieved in space-group prediction if the classes used to train the models comprise more than 100 entries.