search

strawlab / best

Bayesian estimation supersedes the t test
MIT License
145 stars 37 forks source link

example give weird results #3

Closed hfdavidyu closed 7 years ago

hfdavidyu commented 7 years ago

Hello,

when I run example/smart_drug.py, it gives

[-----------------100%-----------------] 110000 of 110000 complete in 11.1 sec Traceback (most recent call last): File "smart_drug.py", line 27, in fig = best.plot.make_figure(M) File "/usr/local/lib/python2.7/site-packages/best/plot.py", line 130, in make_figure posterior_std1 = M.trace('group1_std')[:] File "/usr/local/lib/python2.7/site-packages/pymc/Model.py", line 834, in trace return self.db.trace(name, chain) File "/usr/local/lib/python2.7/site-packages/pymc/database/base.py", line 373, in trace trace = copy.copy(self._traces[name]) KeyError: 'group1_std'

and if I do the M.summary(), it shows the correct means (~ 101) but huge SD (~ hundreds). I am using python 2.7 on OSX. I am wondering what should I do to fix/local the problem?

Sincerely, David Yu

astraw commented 7 years ago

Thanks for the report. I recently discovered that this package has been subsumed into pymc3 (see here). As I think they will do a better job of maintaining the implementation, I intend to put this project into an archival mode.

I am therefore referring you to the implementation in pymc3 and closing this issue.