Issue #12 resolved
As it is now, the prediction is based on a lm model. But Benoit has shown that we did not need to and we can use the original formula. Worth changing if we migrate to sgcca algo in the near future? To be discussed at next meeting.