For readers keeping track, we are currently on the fourth iteration of our Crowd Calendar methodology, which has evolved as we have gotten more data and better technology. In the beginning, the only information we had was data collected by the Testa family doing wait time loops and writing down their observations (seems like 100 years ago). When the Lines app was introduced, however, we began getting hundreds of wait times submitted daily by users.
Today, we get thousands of wait times submitted each and every day. Gifted with all of this wait time data, we can validate and evaluate our predictions and methodologies. Well, it is about time to make some new tweaks, which we plan on implementing for 2017. The reason we are bringing this up now is so users are aware and so we can get feedback. Without our users we wouldn’t be able to do this, so we want to make sure we’re giving you the best product we can.
The Touring Plans Crowd Calendar is based on wait times at specific attractions at each park. Yes, simply using that day’s attendance would be better–and we would love to use it, but Disney simply does not release daily park attendance. The reason we decided on attraction wait times is that they are quantifiable and predictions can be verified.
This method is not perfect, though. For instance, the Disney Hollywood Studios changes over the last couple of years have skewed that park’s crowd levels. There are fewer operating attractions at the Studios right now, and the ones that remain tend to be very popular (Rock ‘n Roller Coaster, Toy Story Midway Mania, et al). Therefore, the wait times have increased, but not because there are more people in the park, just that there is less for them to do once they get there.
The way we have designed the current distribution of our crowd levels is to have half of each year’s days with crowd levels between 1 and 5, and the other days with crowd levels between 6 and 10. With the current settings, there are no days with a crowd level of 1, and very few days with a crowd level of 2. The way we plan on adjusting our settings means that some of the “3” days will become “1” or “2” days. Other parks’ crowd levels are also currently skewed, but not as significant has Hollywood Studios.
The way we go from attraction predictions to crowd levels is a little bit science and a little bit art. When all attraction prediction are very low or very high it is “easy” to score those days as a “1” or “10” respectively. The artistic part comes when selecting the attractions that best represent a park’s crowd levels. We avoid including a new attraction since elevated crowd levels are partly due to its popularity, not the park’s crowd level. The Magic Kingdom has the most attraction and gives us the most flexibility in selecting attractions.
We score each attraction on a scale of 1 to 10 based on the daily average wait time. For Crowd Calendar 4.0 we use wait times between 10am and 5pm as the evaluation period, which was selected because all parks are open between 10am and 5pm. One of the adjustments that we are planning on making is starting the evaluation window to 11am because the 10 o’clock hour is not a good indicator of the entire day’s crowd level–even on crowded days the mornings can have low crowds. By using 11am to 5pm as the evaluation time period we see more variation between days. We don’t use times after 5pm because guests start to leave and are beginning to spend time going to dinner.
Here is a table showing the Magic Kingdom attractions. The list is ordered by the range of daily average wait times with the ranges calculated by dropping the top 5% and bottom 5% days. The median is the daily average wait time where half the days are above and half are below. The attractions that are in bold font are the current attractions used in the Magic Kingdom crowd level.
|Big Thunder Mountain||47||42|
|7 Dwarfs Train||43||93|
|Under the Sea||35||36|
|Princess Hall: Cinderella/Rapunz||34||55|
|Peter Pan’s Flight||33||49|
|Town Sq Mickey||27||29|
|Winnie the Pooh||26||38|
|it’s a small world||23||19|
|Pirates of Caribbean||23||21|
|Enchanted Tales with Belle||21||31|
|Princess Hall: Frozen||17||60|
|Mad Tea Party||13||17|
|Town Sq Tink||7||19|
When selecting attractions to include the crowd calendars, we want a broad range attractions and attractions with a large variation in wait times. Enough time has passed now that we can add the New Fantasyland attractions. Even with Splash Mountain having the largest range of wait times, its wait times are a better prediction of the weather than the crowd levels. Princess Hall changes are coming up so we will not include them right now.
We have empirical data showing that wait times are up, but at Disney’s May 10, 2016, Q2 Earnings Release, Disney said that attendance was down. At this point, we have to assume the higher wait times are the new normal. With the higher wait times and new evaluation window, we have to change the thresholds for each attractions’ crowd level.
The next adjustment we are considering is how we combine attractions’ crowd levels to park crowd level. The Crowd Calendar 4.0 method was to drop the highest and lowest crowd levels and take the average of the remaining attractions’ crowd levels. The new idea is to add an additional step: If majority (50% or more) attractions have the same crowd level, that will be the park crowd level. This change would only affect 30% of the park/days but does increase the number of days with extreme crowd levels.
Thoughts? Suggestions? Questions? Let us know below.