possible PR for #1314 extras and #1528 rebased on current master
unfortunately during cherrypicking one author got switched
isn't this just (resid**2).sum()
instead of the nobs_i, nobs_i outer product
ignore this, residsq is used in the next line
general candidate for conistent renaming across models start_params
but not consistent in other models either
params_mean postfix ?
another term that we start to see more across models, and haven't settled on a name yet.
no reason to change here again, before we haven't settled on a name
Changed covstruct argument to cov_struct
Improved one line of linear algebra
Added a bit to the release announcement for GEE
Remove local test code
Changed covstruct to cov_struct
Doc string fix
Added diagnostic plot for covariance structure
Rename the plot function
Refactored nominal and ordinal code, added plot function for ordinal
rebased again on current master,
should be ready for green button merge when TravisCI reports back.
(The nice thing about isolated imports: I haven't yet recompiled my working checkout for the new cython code, and everything else still works, so far)
Coverage remained the same when pulling 23ead84 on josef-pkt:gee_maint_1528_rebased into cd7f0d0 on statsmodels:master.
I forgot the commit code in the merge message again