QOL improvements

[marcdotson]

This smooths over some of the rough edges and provides needed quality-of-life improvements after merging the previous three branches, including:

• Rewriting scripts to rely on and work with 01_control-file.R.
• Combined simulating data and cleaning empirical data into a single 02_data-prep.R.
• Using a single simulated data file for a variety of ensembles sizes.
• Now simulating data, running the HMNL, and running the competing models only run if it already hasn't been run (i.e., these all only need to be produced/run once for a variety of ensemble sizes). Thus we can change nmember in the control file and easily run it without inadvertently mixing up simulated data files or re-running what we don't need to re-run.
• Ensembles of various sizes are now drawn randomly from the 2000 provided ensembles in a given simulated dataset.
• Produces and append a single ensemble-fit file for a given simulated data and nmember with an updated naming scheme that matches data.
• Including both Gamma and Omega means and 100 log_lik draws in ensemble-fit.
• The ensemble matrices and ensemble weights are now includes as part of the single ensemble-fit object for the given simulated data and nmember.

[marcdotson]

FYI, @RogerOverNOut there was one mistake in predictive_fit_hmnl() that I believe I fixed correctly. The function was still referencing Umat_gammadraws, but looking back at the previous version of the function it looks like that was all changed to Umat when we went to posterior parameter means only. The change is implemented with the old lines of code commented out.

[RogerOverNOut]

Nice catch! that would all be correct.

On Wed, Feb 17, 2021 at 3:08 AM Marc Dotson notifications@github.com wrote:

FYI, @RogerOverNOut https://github.com/RogerOverNOut there was one mistake in predictive_fit_hmnl() that I believe I fixed correctly. The function was still referencing Umat_gammadraws, but looking back at the previous version of the function it looks like that was all changed to Umat when we went to posterior parameter means only. The change is implemented with the old lines of code commented out.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/marcdotson/conjoint-ensembles/pull/61#issuecomment-780381233, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHEP55D24F3VFEXB2U6IJLTS7N2PRANCNFSM4XXUGFNQ .