Fun-inria MOOC scikit-learn Exervice M4.04: Different outputs for linear regression coefficients

fun_linear_models_ex_04.ipynb

my_linear_models_ex_04.ipynb

Hi,

Here are the two files reproducing the issue, that we talked about in the forum, about exact same coefficients between LinearRegression and Ridge on my machine.

The first file fun_linear_models_ex_04.ipynb is the code run from the fun-inria server and the second my_linear_models_ex_04.ipynb run from my local machine.

Note that I’ve added a cell for computing the coefficients by hand.
I had an error raised from the fun-inria server, as the determinant is computed at exactly 0, however the `LinearRegression’ still provides results without explicit warning (although we can understand that the coefficients are unreasonable).
On the other hand, the determinant is computed at approximately 0 on my machine and so I could proceed and got somewhat reasonable coefficients.

At the end of the day, I get the same coefficients between LineaarRegression and Ridge (regularized) on my machine.