Sidewalk alignment #189
Sidewalk alignment #189
Conversation
|
|
|
|
| sorted_local_indices = np.argsort(maybe_members_rank_value) | ||
| sorted_global_indices = \ | ||
| group_maybe_indices[sorted_local_indices] | ||
| if method == 'sidewalk': |
|
|
|
In its current form, the need is not really respected/does not really makes sense (AFAICT). However, it should be possible to simply scale the score/probabilities to achieve the "needed" figures, even if need is N dimensional. |
| "arguments") | ||
| elif method == 'sidewalk': | ||
| raise Exception("If alignment on sum of score is wanted " | ||
| "then use align_abs") |
There was a problem hiding this comment.
It took me several reads to understand what this means. Any idea to make this clearer?
benjello
force-pushed
the
sidewalk_alignment
branch
from
a2efb63
to
cfafa58
Compare
|
|
|
removing small.h5 is a bit harsh. It either needs to stay (but please do not commit it when it's changed), OR run import.yml before you run simulation.yml (and the variants which also use small.h5) |
|
|
benjello
force-pushed
the
sidewalk_alignment
branch
from
ac3cbdc
to
8537823
Compare
|
|
| @@ -179,8 +202,8 @@ def collect_variables(self): | |||
| # in this case, it's tricky | |||
| return set() | |||
|
|
|||
| def _eval_need(self, context, need, expressions, possible_values, | |||
| expressions_context=None): | |||
| def _eval_need(self, context, need, expressions, possible_values, expressions_context=None): | |||
There was a problem hiding this comment.
could you revert this line change?
|
This looks mostly good to me. I just wonder what is wrong about the test using values from .csv files? Anyway, I will not have time for a in-depth review & merge for a couple of weeks. As for the default values PR, could you provide documentation and add a mention in the changelog at doc/usersguide/source/changes/version_0_11.rst.inc |
|
|
Run demo_import before running examples tests
|
|
|
|
|
|
|
Merged! Thanks a lot! |
I resubmit a PR similar to #188 after taking care of some of the issues raised by @gdementen
Remaining potential problems are: