this add more test for parts of the result and for sparse.svds to get more information about the failing test in #1850
result will be visible in pythonxy 32-bit Ubuntu
TST: corr_nearest_factor, add test for sparse svds, objective values
@kshedden this is another try to get at the test failure in corr_nearest_factor
My main guess right now is the starting value or the initial stepsize.
I think adding a start_factor=None keyword would be helpful, unless you think there is no usecase for it.
Coverage increased (+0.01%) when pulling 5ec9fc5 on josef-pkt:corr_nearest_factor_1850 into 20c59d2 on statsmodels:master.
merging, more when the test results are in
see #1850 for continuation