Iain Carmichael

some cases missing opinion files

1

The following cases are missing opinion files when we do bulk download. The few I checked were from 2016. Pretty sure this is CL, not us (but not positive). [3177214,...

compare citations from SCOTUS vs. other courts ## \- plot in-deg|jurisdiction vs. time for jurisdiction = SCOTUS, CA1, CA2, etc > The analysis also shows that measuring precedent using citation...

Assume that igraph graph g has the name attribute. If the name attribute is an integer then I had problems using `g.vs.find(name = n)`.

The pyglmnet solvers give different answers and neither solver agrees with sklearn

3

The pyglment package gives different estimated coefficients for linear regression depending on which solver you select (e.g. cdfast or batch-gradient). Neither solver agrees with sklearn (I believe they should agree!)...

https://github.com/yaglm/yaglm/blob/cefbfc85d96c3c3b187be7df290ba9df9b7b4fc4/yaglm/infer/Inferencer.py#L129

e.g. https://cran.r-project.org/web/packages/glmnet/vignettes/relax.pdf

Running the following code will use an adaptive Lasso, but ignore the weights (similarly if a non-convex penalty is set). This is highly confusing to the user! ```python GlmCV(penalty=Lasso(flavor=Adaptive(), weights=weights))...

The raw materials should be there for this in the opt/cvxpy backends, this is mostly a matter of writing the API.

Come up with a nice way to set `scorer=[scorer1, scorer2, ...]`

Lets not require users to install cvxpy.