Mainak Jas

I think forking might be the easiest. But maybe @pavanramkumar should confirm first if he has any local changes that have not been pushed?

It's a bit hard to help based on this. Could you perhaps provide a full script that reproduces the problem and share a small part of your data along with...

As I said above, please provide a full reproducible script that I can copy-paste. Happy to take it from there :) Thanks!

You have set the learning rate too high, set it to 1e-8 and it should work

The shape of beta depends on the shape of your input data. There is no truncation per se, but if you have too many values in an array, it may...

great, I can replicate now :) I edited your script to remove the netcdf part as it's not needed. I need to ponder on this a bit. In the meanwhile,...

> you are using alpha=1.0 which is equivalent to Lasso regression (L1 penalty). i would not advise this with only 4 predictors in X. Lasso should be used when you...

@titipata do you have time to investigate what is going on?

ummm ... do you mind sharing your data and a small snippet of your code?

this is standard way to invert a matrix, it's the moore-penrose pseudoinverse. Why do you see a problem with this?