2022 Crowd Calendar Retrospective: TouringPlans vs. Disney

It’s the beginning of a new year, and that means it’s a perfect time to start making fun of Disney with data (y’know, again). One of our most popular articles from this time last year was a comparison of “crowd predictions” from Disney and TouringPlans. The basic premise is this: Disney gives us some insight into what they expect crowds to be like by fluctuating their ticket price. Higher ticket prices = higher expected crowds, and lower ticket prices = lower expected crowds. So now that 2022 is over, we can go back and compare actual results with what Disney “predicted” and with what TouringPlans predicted.

All of the Asterisks and Fine Print

As a data nerd, I have to admit that there is a lot of hand-waving behind what appears to be some fancy math. And I wouldn’t be doing my job if I didn’t point out the hand-waving so that we can have some more fun with the results without arguing over their statistical validity, etc.

• First, Disney fluctuates the ticket prices in order to manipulate crowds. Technically, if ticket prices go way up, and crowds are lower, they have in some ways accomplished their mission. One of their goals is to even out crowds throughout the year, so perhaps a more fair fight would be comparing our predictions to all 5s, 6s, and 7s for what Disney “hopes for”.
• TouringPlans predicts and records actual crowd levels based solely on wait times. Disney could have high ticket prices for, say, opening day of the Food & Wine Festival, because they expect large crowds. Those large crowds could descend upon the park. But if they’re all just eating and drinking and roaming and not getting into attraction lines, the actual crowd level we record won’t be all that high.
• Disney gets to control operations. This means that, if they need to (or want to), they can increase or decrease capacity, which in turn affects wait times, and therefore crowd levels. So this makes things slightly less “fair” because TouringPlans can’t do the same thing.

Data Preparation

Before I could compare results, I had to do some data tweaking. Comparing TouringPlans predictions to actual crowd levels is easy. They’re both on a scale of 1 to 10. We define how both are calculated. The Disney piece is more difficult. They had 18 different 1-day ticket prices for 2022, and they weren’t distributed like the nice bell curve that TouringPlans predictions use.

Thankfully, I can bucket the 18 different price points to mirror the TouringPlans distribution as closely as possible. As you can see above, the results aren’t perfect. And that’s fine. Based on Disney’s ticket pricing, they were expecting more very-low crowd days than Touring Plans, and fewer low-moderate (4 and 5) crowd days than TouringPlans. Everything else was roughly equal, other than maybe the crowd level 10 days.

How Accurate Were the Predictions?

Before we get to the big fancy summary, let’s look at some physically bigger, and also probably fancier calendar-style heat maps. These calendars show the difference between the actual crowd level and the predicted crowd level. That means overpredictions are negative numbers (and will show up as purple) and underpredictions are positive numbers (and will show up as orange). Days that are off by only one level are considered to be “good enough” and aren’t colored in these charts.

I like how the visuals can tell us a strong story here. Crowds were higher than ticket prices would suggest in January, February, and most of the summer. And they were significantly lower than ticket prices would suggest in October, November and December. Plus, the dynamically higher weekend pricing in March, April, and May didn’t mean that weekends were more crowded. Saturdays were lower than ticket prices would have predicted.

The TouringPlans calendar is interesting too. Many, many more non-colored cells. But a few of the colors stick out. Specifically, TouringPlans predicted many more crowds to stick around in early January than what actually happened. Similarly, March was somewhat less crowded than expected. June and July, just like for Disney, were slightly more crowded than expected. And then in September and October, a really weird pattern emerges. Well, a weird pattern if you haven’t read any of my data dumps or continual pleas to go to Magic Kingdom on party days. See, there were Halloween parties almost every Tuesday and Thursday. And those party days forced the crowds to Magic Kingdom on Wednesdays. Our original predictions didn’t handle that behavior well. Good thing if you pay attention to me, you knew to expect it 😉

How Do the Predictions Compare to Each Other?

These next graphs should look eerily familiar to anyone following along with our weekly Disney Data Dump:

The results here speak for themselves. But I can’t help myself, so I’m going to speak for them too. Disney’s “predictions” were within 1 crowd level just over half of the time, while TouringPlans hit that same mark 75% of the time. Even more importantly, Disney was off by 3 or more crowd levels 21% of the time, and TouringPlans was only off by that much 4% of the time.

It’s also important to know that if we compare this to 2021 performance, TouringPlans did significantly better this year. Last year, they were only within 1 crowd level 66% of the time. But … give Disney credit where credit is due. In 2021, they were only within 1 crowd level 28% of the time. So their “accuracy” also improved significantly.

What are your thoughts about these results? Is ticket price a serviceable corollary for what Disney thinks crowds will be like? Do you want to see more retrospectives on our predictions from 2022? Let us know in the comments!