search

OahuMPO / archived_travel_model

Official Travel Model for the Oahu Metropolitan Planning Organization
http://wsp-sag.github.io/OMPOTravelModel/
0 stars 0 forks source link

Update visitor units by TAZ #52

Closed dkyleward closed 3 years ago

dkyleward commented 4 years ago

Lori provided the attached file, which should be used to improve the allocation of visitor units by TAZ. She raised concern about their distribution in her email on 10/12/2020. TAZ2040 (150930).zip

dkyleward commented 3 years ago

I have updated the distributions. The previous distribution was so far off it must have been based on outdated TAZ IDs. The maps below show hotel locations. Rental condos follow the same pattern but are not shown.

Previous distribution of hotels:

hr_old

New distribution:

hr_new

dkyleward commented 3 years ago

@rickdonnelly I have updated the visitor info like we discussed. I reverted the RC field back to how it was before. I maintained Lori's distribution for the HR field but scaled it up to 39,864 per DEBDT. I'm going to close this issue again, but open it back up and message me if anything else needs to be done.

rickdonnelly commented 3 years ago

I led you astray, so this isn't correct. As I noted to Lori yesterday the DBEDT 2045 target for total visitor accommodation units (VAUs) is 39,864. But in diving through the HTA Visitor Plant Inventory reports (done by DBEDT prior to 2009) I've zeroed in on how they classify accommodation types. Only 69.5 percent of the VAUs in 2020 were hotel rooms, and if I run a straight line through those points we can assert that 63.1 percent of total VAUs in 2045 will be hotel rooms. So the sum of our HR field should be 39,864 * 0.631 = 25,154 units, and the sum of our RC field should be 39,864 - HR = 39,864 - 25,154 = 14,710 units. So we'll need to rescale those two variables to match these totals.

There is a slight rub in that the visitor model documentation suggests that HR was assumed to equal DBEDT's total VAU during estimation. But I cannot find code that verifies that yet. Moreover, my testing of the visitor model suggests that the effects of doing so doesn't change the outcome, as the visitor units are attraction variables rather than size terms in the destination choice model. But still, we should be consistent with the definitions used in the Visitor Plant Inventories. Thus, until we overhaul the model our RC field is really more than just residential condos, for it will include all accommodation types other than hotel rooms.

dkyleward commented 3 years ago

Done!