Scikit-learn high level interface. I think that this should be the default interface for qml and that we should hide more low level stuff a bit away.
The implementation is about 3/4 done and is ready for beta testing. 'examples/qmlearn.py' shows how to use it.
I had to restructure and modify quite a lot of files, but only major changes was made in the qmlearn submodule.
It is hardcoded to predict energies at the moment, but energies can be replaced with any molecular property. Atomic properties is not supported yet.
Scikit-learn high level interface. I think that this should be the default interface for qml and that we should hide more low level stuff a bit away.
The implementation is about 3/4 done and is ready for beta testing. 'examples/qmlearn.py' shows how to use it. I had to restructure and modify quite a lot of files, but only major changes was made in the qmlearn submodule.
It is hardcoded to predict energies at the moment, but energies can be replaced with any molecular property. Atomic properties is not supported yet.
Tests are on the todo list.