I’ve been a customer of Touring Plans for several trips over the past 9 years. I’m also a Software Engineer who has some experience in data science and have, for fun, studied a lot of the data TP provides (they publish an awful lot of their data, including historical data and how they did). And their success rate is very high. It’s been spot on for every one of my trips, and you can look back and see actual crowd levels vs predicted for every day since 2011.
Lauren, my angst is not so much with the fact that I am going to have to wait an additional 30 minutes for Small World. The problem is that TP is selling a faulty product - plain and simple. There is a certain audacity in providing hour by hour predictions by park five months (or more) in advance when the accuracy of those predictions is +/- 100%. In my business (writing software), we call that a WAG. And, a WAG is not science - data or otherwise.
The +/- 100% isn’t really accurate. Going from a 5 to a 10 doesn’t mean that it’s double the crowds. It is closer to (although not exactly) percentiles, where a 10 is in the 90th percentile of days, a 5 would be the 40th percentile. Again, it’s not quite that but it is closer to that than it is to double the crowds. So saying it’s a +/- 100% isn’t accurate.
It’s also not right to call it a WAG. In data science, a change in inputs can have a high effect on the model. It’s important to remember that what TP does is a forecast. It’s based on past history, numerous other factors (e.g. state of the economy, school schedule across the US, etc.). But when one of those inputs unexpectedly changes, it throws off the model and they have to recalibrate.
I’ll give you an example. Last winter, Disney tried an experiment and cut staffing. They effectively ran many rides at 50% capacity. This was something they had never done before. Wait times skyrocketed. There is no way anyone or any model could have predicted this. (By the way, the model is factoring in that as a risk for this winter too.).
It would be a WAG if the model was consistently wrong and consistently adjusting in this way. But my experience and history shows that this isn’t the case ordinarily. This just happens to be a bad perfect storm. I’m sorry it affected you (FWIW, it affects me too as I’m going President’s Day week - but I know enough to not let it bother me).
In the end, those days could be anything between 4 and 12. Since we don’t have reliable estimates, we have no way of knowing. So, we’ll roll with it. (Needless to say, I won’t be renewing my subscription.)
Mike, If the factors that go into the models are changing and there is no way to validate them, then don’t publish the estimates since they don’t provide value. And, I appreciate the encouragement. But, if you honestly believe that a schedule will turn a 10 into a 5 then I’m guessing that you have never experienced a 10. I’ve been in WDW during Christmas when it was hard to breath in those parks. Making a plan for those days is futile at best.
This isn’t close to a Christmas 10. You can view the actual peak wait time predictions for your days and compare them to Christmas week. They aren’t anywhere near the same. I’ve been on non-Christmas 10 days. With the Touring Plan software, I never waited more than 30 min for anything.