Mattia Gollub

icon location Director of Machine Learning @ JURA Bio

Results 10 comments of Mattia Gollub

Resolve Km entries to actual reference

I just encountered the same issue. When you say that it cannot be fixed do you mean that there is not sufficient information or that it would be difficult with...

Resolve Km entries to actual reference

Ok, it's not as easy as I first thought but the function below works for me. In cases where at least one protein has a comment and at least one...

Nonreproducible predictInteval results mixing observed and unobserved levels.

I changed the title to reflect the main issue, which is the prediction of different intervals with multiple calls for the same query. In my case I want to use...

Nonreproducible predictInteval results mixing observed and unobserved levels.

Here is an example. For clarity: grades come from different tests that each student took during the year, and for whatever reason some grades/pupils have been "lost": ``` library("lme4") library("merTools")...

Nonreproducible predictInteval results mixing observed and unobserved levels.

I made quite a bit of confusion with the comments about the sources of variation. Using the same example and assuming that tests are noisy observations of the skills of...

Nonreproducible predictInteval results mixing observed and unobserved levels.

That sounds very helpful, thanks a lot!

Resolve Km entries to actual reference

Glad it helps! And I noticed the issue of exceptions as well. For example the code above needs to split comments on `>;` instead of `;` and then add back...

Docs: how to handle when CVXPY doesn't recognize an expression as DCP/DQCP

Thank you for the clarification, indeed I mixed up the properties of functions and atoms. I'm still convinced that `f` is quasiconvex and I could model it in DQCP using...

Docs: how to handle when CVXPY doesn't recognize an expression as DCP/DQCP

Thanks everyone for the answers. Ultimately I concluded that I cannot use DQCP because I need to work with `cp.sum(f)`, which is not quasiconvex. My last resort is to use...

Extracting core functionality into a separate package

Hi everyone, Nice to see such an active discussion! I'll add my view as well: - In general, I like the idea of a solver-free model class. Before settling for...