Project Meeting 2017.03.23 - bellevuewa/BKRCast GitHub Wiki
Employment
ESD still considering request
Sampling plan
Nagendra shared our in-progress analysis of different zone-based sampling plans
Have run 8 scenarios with 10 seeds each
Current results across different metrics suggest a BKR 400%, outside BKR 40% sampling plan
This ~70% of total regional HHs plan is close to the optimal compromise between stability and runtime
Note skims and shadow prices are held constant across tests
Will make a few more runs to better understand results
Then will turn to analyzing destination choice sampling size plans and then feedback
Documentation
RSG continues to complete / clean-up the wiki documentation
@Nagendra has some additional revisions to make over the next week or so
SoundCast improvements
Since BKRCast was created, PSRC has improved the external model, airport model, JBLM estimates, and changed the way they seed skims and do shadow pricing
We won't integrate any of these changes right now since we want to do calibration first and these improvements won't impact BKR all that much anyway
BKR may do later with PSRC help
Calibration
@Nagendra to commit, document in wiki, and share with @Hu the calibration summaries scripts
@Hu to run them on the in-house ESD-based model and share summaries with RSG