I think there has been some changes in MLJLinearModel defaults which required some attention here.
For the ElasticNet model, I couldn't quickly find parameters that did not lead to convergence issues. This is possibly tricky here in any case because the data set is so small. So I've opted to just drop that last model.
However, the "zero coefficient" outcome for Lasso has been fixed.
I think there has been some changes in MLJLinearModel defaults which required some attention here.
For the ElasticNet model, I couldn't quickly find parameters that did not lead to convergence issues. This is possibly tricky here in any case because the data set is so small. So I've opted to just drop that last model.
However, the "zero coefficient" outcome for Lasso has been fixed.