Unofficial Guide Crowd Variables

by

  • Facebook
  • Twitter
  • Google Plus
  • Pinterest

The latest model I’m using has four variables as input: historical hotel occupancy rate for I-Drive and LBV hotels, number of hours the MK is open during the week, trips booked in the Unofficial Guide’s trip planner, and a measure of WDW resort room availability (or lack thereof) in Expedia, plus Swan & Dolphin and Hilton in DTD. If we compare the average of those four things with the known wait times for the past 14 days we have data for, here’s the accuracy:

Matches exactly: 4 out of 14
Predicted level is 1 more than actual: 3 out of 14
Predicted level is 1 less than actual: 6 out of 13
Predicted level is off by 2 or more: 1 out of 13 (predicted 7, was 5 – might have been weather)

The underestimation is interesting. I’m going to work with these numbers some more to see if the accuracy can be improved. If we can get it to where we predict either a specific number (e.g., 6) or a range (e.g., 6-7), that might be a good first step. In any event, here’s the average of those four numbers for the next few days

Fri Apr 22
8
Sat Apr 23
8
Sun Apr 24
6
Mon Apr 25
7
Tue Apr 26
7
Wed Apr 27
6
Thu Apr 28
6
Fri Apr 29
6
Sat Apr 30
5
Sun May 1
5
Mon May 2
5
Tue May 3
5
Wed May 4
5
Thu May 5
5
Fri May 6
5
Sat May 7
6
Sun May 8
6

Those numbers use the old Unofficial Guide scale, not the new. It might also be interesting to see if we use the weekly hours as a base, and the daily hours as a more precise estimate. I don’t think it’d work on Grad Nights, or any time the MK has a special event, so it’s tough to say right now. Just a thought.

  • Facebook
  • Twitter
  • Google Plus
  • Pinterest
Posted on April 21, 2005
Comments are closed.