bgoodri

I think that is what is supposed to happen.

The Stan code distinguishes between the `has_weights=1` and `has_weights=0` cases, as in https://github.com/stan-dev/rstanarm/blob/master/src/stan_files/continuous.stan#L203 So, I would expect a dataset that has been collapsed to the unique values but has a...

I will but I'm not sure when. Both rstan and Columbia are in a state of flux ATM. On Mon, Mar 9, 2020 at 2:46 PM Eric Potash wrote: >...

Or a zipped file containing the .csv files produced by Stan. See the section entitled "Compression" at help(unzip). Also, see https://groups.google.com/forum/#!searchin/stan-users/read_stan_csv/stan-users/WMfp61h8m0A/iIKwE7toKf8J Also, some guy had more success using fread function...

Is this tantamount to a `posterior_epred` method for survival models?

Enforce it even if Andrew disagrees. On Wed, Feb 1, 2017 at 12:54 PM, Jonah Gabry wrote: > @sakrejda @paul-buerkner > @imadmali > Thanks for the feedback. Glad you agree....

The short answer is that with `stan_gamm4`, the design matrix for the smooth function has been transformed by a SVD so that its columns are orthogonal and hence the independent...

The stochastic representation of the multivariate normal is in terms of the Cholesky factor of the covariance matrix. I guess you could do something with the inverse of a Cholesky...

Most of the time (i.e. all of the time as far as rstanarm is concerned) you would want to do the multivariate version of this rather than the marginal &...

It isn't so complicated: If you can construct the precision matrix, you can use the inverse of its Cholesky factor in an uncentered parameterization of the coefficients. A similar thing...