Closed ghost closed 4 years ago
There is a bug with the convergence at the moment. See #275
This would be too verbose in my opinion. We should instead fix #275 first. Any help there is welcome
Merging #277 into master will increase coverage by
0.68%
. The diff coverage is100%
.
@@ Coverage Diff @@
## master #277 +/- ##
==========================================
+ Coverage 74.15% 74.84% +0.68%
==========================================
Files 4 4
Lines 623 640 +17
Branches 127 127
==========================================
+ Hits 462 479 +17
Misses 125 125
Partials 36 36
Impacted Files | Coverage Δ | |
---|---|---|
pyglmnet/pyglmnet.py | 80.93% <100%> (+0.65%) |
:arrow_up: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 8b15c58...d936152. Read the comment docs.
@jasmainak this is probably ready for another look.
I rebased to the current master and made sure it only logs each iteration's convergence when the logger is set to DEBUG
- it's now silent at the default INFO
level. It may also address #287.
This records the number of iterations within the GLM object, and whether or not the
fit()
iterations stopped because it converged or because it hitmax_iter
. It also logs a warning ifmax_iter
is reached before convergence.In the current pyglmnet master branch, the only indication of convergence is in the logs, so this change makes it easier to check convergence programmatically.
It passes
nosetests . --with-coverage
locally, andgit diff main/master -U0 | flake8 --diff -
flags no pep8 violations.