News Star Wars Galaxy's Edge Disney's Hollywood Studios opening date

mikejs78

mikejs78

Premium Member
TheDude23 said:
This sentence cracked me up. I love all of our levels of Disney fandom to where that sentence sounds completely normal and reasonable. :)
Click to expand...
You laugh, but @lentesta has done just this before. Correct me if I am wrong Len, but you sent people to count people in lines during winter of 2018 and found that capacity was substantially reduced.
 
winstongator

winstongator

Well-Known Member
lentesta said:
Cool. Let me know what you come up with.
Click to expand...
358322

Here's my histogram. I used a scatter plot as I used non-uniform bins. As the standard deviation predicted, about 2/3rds of the numbers are between -10 & +10 minutes for error. There is a skew to very slightly overestimating waits, as most would predict. Highest frequency bin is -5 to -10 min error (5 to 10 minutes less actual wait).

If I used uniform bins, one, I'd need a ton because I want granularity for the very low wait times, two the data points past 30 min or so would be quite small. I'll play around with it though. The other thing I'd like to see is wait time error vs wait time. If wait times are 20 min or less, it's hard to screw those up much, and people don't really care about them.

The other thing to look at is how does AK alone look, perhaps by year?
 
winstongator

winstongator

Well-Known Member
lentesta said:
Cool. Let me know what you come up with.
Click to expand...
358324

Here is wait time error vs. actual wait. It's bounded on the y-axis by each x-axis value - your wait time error can't be more than your actual wait - there is no bound on the negative side. I don't know if this plot shows anything useful.

Google sheets does not like this file...
 
winstongator

winstongator

Well-Known Member
lentesta said:
Cool. Let me know what you come up with.
Click to expand...
Google sheets, or my mac, does not like what I'm doing - cutting a large number of rows. I think it's sheets as I'm deleting the rows, not cutting them (to the clipboard...) My version of excel was not able to import the file either. That makes analyzing the data slower than I'd like.

I haven't used shell scripts like you have in the GitHub repository before. Debating whether to tackle trying that, building a new script in python or continuing to wrestle with sheets and excel.
 
winstongator

winstongator

Well-Known Member
lentesta said:
Cool. Let me know what you come up with.
Click to expand...
358349

I started looking at just AK. There are 10 attractions listed. Tough to be a bug is very accurate. Whirl and Triceratops are more accurate. Everest, while error is up from 2018, is actually down from 2017 - so call that even. Kali River Rapids is a very small dataset for the winter (6 points). That's half of the attractions close. Outpost is a small dataset too. Navi has been pretty consistent in its error. That's 7/10. We've then got FoP, Kilimanjaro Safari and Dinosaur. The graph above is averaged FoP waits (red), and wait time error (blue). Error has a pretty large negative shift early in the year before trending towards zero for the summer/early fall, then back down slightly late in the year. If you look at just Jan & Feb 2018, the error is actually bigger: -36 min vs. -32 min! The safari is seeing more error in the waits, but also longer waits. The difference does not go away like FoP when you look at just the same months. The variation in the wait time error is also up which could be a signal that they are somewhat just off by a percentage of the wait time.

I would need to look at the data more closely, but I would guess that you would have bigger average errors when you are looking at bigger wait times. If you're off 5% on 20 min waits, that's 1 min error, 5% on 60 min waits is 3 min. Actual percent errors are different.

For Dinosaur, if you look at just the same range of dates in 2018 as 2019, the average error is -6.8 vs. -6. The shift from that to -8.1 is less. These average differences still seem pretty small.

The biggest effect seems to be for the Safari, and a big part of that is driven by its increased wait times.

Looking at the data, I have a lot more appreciation for Disney's posted waits. They are pretty darn accurate.
 
MagicHappens1971

MagicHappens1971

Well-Known Member
I'm a little late in adding this in, but Hogsmede/Wizarding World at Islands of Adventure even years after it opened had to issue return times to guests because they couldn't fit enough people in. I wouldn't be surprised if Disney went this route for Galaxy's Edge.
 
lentesta

lentesta

Premium Member
winstongator said:
View attachment 358349
I started looking at just AK. There are 10 attractions listed. Tough to be a bug is very accurate. Whirl and Triceratops are more accurate. Everest, while error is up from 2018, is actually down from 2017 - so call that even. Kali River Rapids is a very small dataset for the winter (6 points). That's half of the attractions close. Outpost is a small dataset too. Navi has been pretty consistent in its error. That's 7/10. We've then got FoP, Kilimanjaro Safari and Dinosaur. The graph above is averaged FoP waits (red), and wait time error (blue). Error has a pretty large negative shift early in the year before trending towards zero for the summer/early fall, then back down slightly late in the year. If you look at just Jan & Feb 2018, the error is actually bigger: -36 min vs. -32 min! The safari is seeing more error in the waits, but also longer waits. The difference does not go away like FoP when you look at just the same months. The variation in the wait time error is also up which could be a signal that they are somewhat just off by a percentage of the wait time.

I would need to look at the data more closely, but I would guess that you would have bigger average errors when you are looking at bigger wait times. If you're off 5% on 20 min waits, that's 1 min error, 5% on 60 min waits is 3 min. Actual percent errors are different.

For Dinosaur, if you look at just the same range of dates in 2018 as 2019, the average error is -6.8 vs. -6. The shift from that to -8.1 is less. These average differences still seem pretty small.

The biggest effect seems to be for the Safari, and a big part of that is driven by its increased wait times.

Looking at the data, I have a lot more appreciation for Disney's posted waits. They are pretty darn accurate.
Click to expand...
What happens if you restrict wait times to those observed between 10 am and 5 pm?
 
olie64

olie64

Well-Known Member
lentesta

lentesta

Premium Member
winstongator said:
I realize that google sheets is not an adequate tool for data analysis.
Click to expand...

Heh.

You probably don't want to average the posted and actual waits to determine accuracy, because being off by (say) 25 minutes each way on two samples would result in a net error of 0.

I'd also restrict the wait times to 10 am to 5 pm, which is peak crowd time. You'll have fewer "start-of-day posted waits of 5 / actual waits of 3" pairs, which don't tell you as much.

One thing I haven't done yet is compute the confidence interval of the posteds and actuals. By month and attraction would be interesting. And at different confidence levels.
 
winstongator

winstongator

Well-Known Member
lentesta said:
Heh.

You probably don't want to average the posted and actual waits to determine accuracy, because being off by (say) 25 minutes each way on two samples would result in a net error of 0.

I'd also restrict the wait times to 10 am to 5 pm, which is peak crowd time. You'll have fewer "start-of-day posted waits of 5 / actual waits of 3" pairs, which don't tell you as much.

One thing I haven't done yet is compute the confidence interval of the posteds and actuals. By month and attraction would be interesting. And at different confidence levels.
Click to expand...
I have been looking at average actual wait time, the standard deviation of that actual wait time, the average error, and the standard deviation of that error. Averaging the error would pick up the bias towards overestimating waits, and the standard deviation of the error would be a measure of how well WDW is estimating the waits.

I noticed a lot of pairs of data, so for the Safari & Everest data I looked at, I hand trimmed those out (ouch). I also only looked at Jan & Feb for the prior years. Findings summarized below. Safari shows a larger error, but less of a change from 2017 than 2018. Also interesting that 4min of that average error is from 7 outlier datapoints. Everest waits seem more accurate recently, and Dinosaur similar to those other years.

My hypothesis: WDW wants to err on the side of a longer posted wait. Then guests are happily surprised when they wait for fewer minutes. As average waits go up, you need more of a bias to ensure that. Also, as wait times become more variable, you need a bigger bias as well. I need to write up a simple example of this to double check myself.

I graphed an average of the Safari & Everest Actual Waits & Posted wait error. You can see a trend in actual waits going up, but it's hard to see a real trend in the errors.

I histogrammed some of the data too. For one observation interval 80% of the measured actual waits were within 10 minutes of the posted wait. Most intervals were 70%. The 2019 interval of the safari was at 60% (it wouldn't take much convincing for me to believe that's an outlier and less indicative of a trend).

I have one vivid memory of a posted wait being much higher than actual. Around this time 3 years ago, we were there with friends. Mine train posted wait was 90-120 min, but I couldn't see anyone past the gem game spot. It was around a 20 min wait. I think it might have been right after a down time for the ride, but it worked out for us.

Safari​
​
average measured wait​
std dev of meas. wait​
avg error​
std dev of error​
2017​
31.8​
24​
-7.8​
16.9​
2018​
42.7​
17​
-5.9​
12.4​
2019​
46.8​
24​
-14​
20.7​

Everest​
​
average measured wait​
std dev of meas. wait​
avg error​
std dev of error​
2017​
20​
18​
-12.5​
17.9​
2018​
27.6​
16​
-15​
23​
2019​
33.2​
15​
-9.3​
13.2​

Dino (less rigorous look at data)​
​
average measured wait​
std dev of meas. wait​
avg error​
std dev of error​
2017​
25.5​
18.4​
-7.1​
23​
2018​
27​
16.4​
-8.6​
14​
2019​
32.9​
19.3​
-8.2​
16.5​

Safari Waits (blue) and Wait error (red)
358622


Everest Waits (blue) and errors (red)
358621
 
Keppyslinger

Keppyslinger

Well-Known Member
I have a trip lined up for Disney in October, and while I understand Galaxies Edge will not be offering Fastpasses, will park capacity be an issue? Like if I have a RnR fastpass am I potentially locked out of the park because so many people are trying to get into GE?
 
nickys

nickys

Premium Member
Keppyslinger said:
I have a trip lined up for Disney in October, and while I understand Galaxies Edge will not be offering Fastpasses, will park capacity be an issue? Like if I have a RnR fastpass am I potentially locked out of the park because so many people are trying to get into GE?
Click to expand...

We don’t really know yet. Until it opens at DL and they see how it works, there are probably several possible ideas as to how it will work at WDW.

My hunch is there will be a separate line for Galaxy’s Edge. But they will need to factor that into the park capacity. But phased closures of DHS are possible, even likely. And in that case, staying onsite, having an ADR and/or FPs should at least give you a decent chance at getting in. Onsite guests are the last to be denied entry.
 

Register on WDWMAGIC. This sidebar will go away, and you'll see fewer ads.

Back
Top Bottom