Feras Saad

> Picking it at random gives the correct distribution, in aggregate, at least -- but only if we use the same generator for every INFER -- but only if we...

Just hit this issue as a bug proper. ``` bdb.execute(''' CREATE POPULATION satellites_p for satellites_t WITH SCHEMA ( GUESS STATTYPES FOR (*) ) ''') bdb.execute(''' CREATE METAMODEL satellites_m FOR satellites_p...

Actually, what broke this behavior was @riastradh-probcomp's fix https://github.com/probcomp/bayeslite/commit/d7c5b3fdb419a5ea2611057b881ea02331d27d2b to this issue, which appears to have caused the regression. Update: git revert d7c5b3fdb419a5ea2611057b881ea02331d27d2b indeed fixed the issue with Latent variables...

@riastradh-probcomp I'm going to revert this commit --- it has completely broken the create schema logic in https://github.com/probcomp/bayeslite/blob/master/src/metamodels/cgpm_metamodel.py#L1043, and refactoring that complex state machine to is likely to be more...

Related, perhas https://github.com/probcomp/bayeslite/issues/18

For those looking for this feature, a simple workaround is to create a table of simulations, and use sql select to apply f to the samples.

The bayeslite shell is not being developed at the moment in favor of jupyter. There is a `.population` magic, not sure what `.describe` for a table would do but is...

``` simulate weight, t1.(estimate * from columns of p1' ' where probability of (mutual information with age < 1) > 0.8)' ' from p1 limit 10') ```

Yes, In addition to the base measure there are also the qualitative invariants that define the statistical type more completely -- for instance a count variable is very different than...

@curlette Even though you specified `k=7` in the MML code defining the number of categories for the random forest, the "training data" in `scorecard_training_data` only contains values `[0,1,2,3,4]`, which means...