idsia/bayesrecon

Source of bayesRecon R package 📈

bayesrecon's Issues

Test warning in .fix_weights

Create a test for the case where .fix_weights returns a warning.

Check entries of S

Check that each entry of S is either 0 or 1. If not, raise ValueError.

Fix schaferStrimmer documentation

The output shape is p x p (instead of n x n)

reconc_gaussian: return Full Cov

return a Full reconciled covariance matrix instead of blocks (upper / bottom)

size = 0 in negbin

If parameter size = 0, raise an error (instead of sampling from a Poisson)

Check if A is hierarchical

Check if A is a hierarchy matrix before running the integer linear programming problem to find H, G.

Check distr

Check that each input distribution string is one of the allowed ones, and raise an error otherwise.
Check that there isn't any continuous distribution that is child of a discrete one.

Fix reconc_BUIS documentation

The required parameters for the negative binomial distribution are mu and size (instead of mu and prob)
The correct string for specifying the negative binomial distribution is negbin (instead of nbinom)

List of strings for "in_type" in reconc_BUIS

Add possibility to pass a list of strings as in_type, each entry can be either "params" or "samples"