Rerun glm-logistic #40
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
|
@MarcoGorelli What's missing here? |
|
I think there's a few errors, but I plan to address them this week EDITdidn't get round to it this week in the end, not sure when I'll have time, sorry |
|
View / edit / reply to this conversation on ReviewNB OriolAbril commented on 2021-03-08T15:36:40Z I would suggest using https://arviz-devs.github.io/arviz/api/generated/arviz.sel_utils.xarray_var_iter.html#arviz.sel_utils.xarray_var_iter. We just modified the return signature so now it's not as convenient as what I did in https://oriolabril.github.io/oriol_unraveled/python/arviz/matplotlib/2020/06/20/plot-trace.html#The-lines-argument, but it is in general much better practice and will work for scalar and multidimensional variables, whereas the summary approach gets tricky with multidim variables.
I am also seeing that it is probably good idea to allow having the |
|
View / edit / reply to this conversation on ReviewNB OriolAbril commented on 2021-03-08T15:36:41Z The labels are wrong here, as it uses plt they are added to the rightmost axes instead of being added on the central one |
|
View / edit / reply to this conversation on ReviewNB OriolAbril commented on 2021-03-08T15:36:41Z not sure if it only happens in reviewnb, but the italics formatting is wrong |
|
View / edit / reply to this conversation on ReviewNB OriolAbril commented on 2021-03-08T15:36:42Z We would have to look at the internals of plot_glm to make sure it won't automatically convert to ndarray, but using xarray here would significantly simplify the code. All broadcasts would go,
It requires initializing the ages linspace as a dataarray instead, but after that broadcasting is automatic.
We can open an issue about that to have someone else do it in a follow-up PR too. |
|
related to #82 |
|
closing as stale for now as I'm never getting round to this and there are other PRs addressing glm notebooks anyway |
* refactor * use keys * sample k
Use inferencedata / ArviZ
This one took 9668.2s on Kaggle kernels with GPU!