Optimal park commando plan

[KeithVH]

Couldn't find anything in search so asking out loud. Has anyone ever done an analysis on optimal path through a given park? For sake of argument, let's limit the discussion to MK. I mean, I've read lots of supposedly logical reasoning and expressions of personal experience about which way to go off of Main Street and which rides to do in which order.

But does anyone have any EMPIRICAL evidence to support a claim that their method is better than any other? The idea is to be the most efficient track through the park while reducing all ride wait times. I'd like to think there IS something of an absolute best way of doing things with a certain set of variables. Those being (at the least):

• Time of year of the visit
• Time of day entering the park
• General weather condition
• Availability/refurb status of all rides
• Wait time trends as related to time of day/time of year

What items am I missing? Heck, you could probably even do this in Excel by scoring impact of the above variables and weighing scores for desirability of getting on any given ride at a given point in time as efficiently a possible and experiencing as many attractions as possible within a set time period. Unfortunately, I lack the hard data required for a sufficiently large enough period of time to even take a stab at this. Even taking a swag at numbers will run me into some observer effect problems and personal bias.

Anyone else, among the geeks here, ever done something like this?

 

[englanddg]

Couldn't find anything in search so asking out loud. Has anyone ever done an analysis on optimal path through a given park? For sake of argument, let's limit the discussion to MK. I mean, I've read lots of supposedly logical reasoning and expressions of personal experience about which way to go off of Main Street and which rides to do in which order.

But does anyone have any EMPIRICAL evidence to support a claim that their method is better than any other? The idea is to be the most efficient track through the park while reducing all ride wait times. I'd like to think there IS something of an absolute best way of doing things with a certain set of variables. Those being (at the least):

• Time of year of the visit
• Time of day entering the park
• General weather condition
• Availability/refurb status of all rides
• Wait time trends as related to time of day/time of year

What items am I missing? Heck, you could probably even do this in Excel by scoring impact of the above variables and weighing scores for desirability of getting on any given ride at a given point in time as efficiently a possible and experiencing as many attractions as possible within a set time period. Unfortunately, I lack the hard data required for a sufficiently large enough period of time to even take a stab at this. Even taking a swag at numbers will run me into some observer effect problems and personal bias.

Anyone else, among the geeks here, ever done something like this?

www.touringplans.com has exactly what you are looking for. And, their methods are quite solid.

 

[Tom]

Couldn't find anything in search so asking out loud. Has anyone ever done an analysis on optimal path through a given park? For sake of argument, let's limit the discussion to MK. I mean, I've read lots of supposedly logical reasoning and expressions of personal experience about which way to go off of Main Street and which rides to do in which order.

But does anyone have any EMPIRICAL evidence to support a claim that their method is better than any other? The idea is to be the most efficient track through the park while reducing all ride wait times. I'd like to think there IS something of an absolute best way of doing things with a certain set of variables. Those being (at the least):

• Time of year of the visit
• Time of day entering the park
• General weather condition
• Availability/refurb status of all rides
• Wait time trends as related to time of day/time of year

What items am I missing? Heck, you could probably even do this in Excel by scoring impact of the above variables and weighing scores for desirability of getting on any given ride at a given point in time as efficiently a possible and experiencing as many attractions as possible within a set time period. Unfortunately, I lack the hard data required for a sufficiently large enough period of time to even take a stab at this. Even taking a swag at numbers will run me into some observer effect problems and personal bias.

Anyone else, among the geeks here, ever done something like this?

Based on your subject, I thought you were asking the best way to tour the parks without any underpants on. My suggestion would be to go during the less sweaty times of the year, walk slowly, and dress otherwise appropriately

But since that's not what you were asking, I second what @englanddg said - go to TouringPlans.com. @lentesta has done all the hard work for you.

 

[FettFan]

*no underwear.

 

[WhatJaneSays]

I’m in the process of trying to determine something related to this. My objective is to complete a WDW46 (http://www.parkeology.com/wdw46-faq) before it becomes WDW46+new rides in such a short time that no one can beat it. Looking for the “optimal path” across 4 parks has been a bit headache inducing.

I actually am using excel in this endeavor! I can’t even guess at how many crowd calendars, wait time charts, and touring plan type things I’ve used at this point. So far just my “final calculations” workbook is at 17 sheets. I’ve picked my optimum dates for the attempt and have planned out 3 tentative schedules. Working on my own time trials is my next step for refinement. (As a single person with FPs and good timing I can now clear Studios in 2 ½ hours.)

. . . however I’ve been trying to make the WDW46 happen since February and I’ve recalculated my plans nearly 2 dozen times. I feel there is no surefire way to know in advance the most expedient path – just a really well educated idea of one.

(attached screen caps of my insanity for hilarity)

 

[lentesta]

Couldn't find anything in search so asking out loud. Has anyone ever done an analysis on optimal path through a given park? For sake of argument, let's limit the discussion to MK. I mean, I've read lots of supposedly logical reasoning and expressions of personal experience about which way to go off of Main Street and which rides to do in which order.

But does anyone have any EMPIRICAL evidence to support a claim that their method is better than any other? The idea is to be the most efficient track through the park while reducing all ride wait times. I'd like to think there IS something of an absolute best way of doing things with a certain set of variables. Those being (at the least):<snip>

Yeah, this was the subject of my Masters thesis, in computer science. I turned in to the software that runs touringplans.com, and I co-author the Unofficial Guide to WDW. The software has been available (for free!) online for years. Sometime in the next couple of weeks we'll optimize our 8 millionth touring plan.

Back to the touring plans - they're really hard to do well, because of the number of permutations you have to consider. A plan with 10 attractions have 10! (10x9x8x7x6x5x4x3x2x1 = ~3.628 million) possible touring plans. 20 attractions have 2.43 x 10^18 possible plans, so you're not going to go through all of those in any reasonable timeframe. That means you have to use heuristics, and that's what my thesis covered.

You'll save 4-5 hours in line on a busy day with a good plan, versus people who don't have a plan.

While it's hard, none of this is magic. It's the application of math and science to what is essentially a closed system of queues.

Thanks for mentioning the site, @englanddg and @Tom

 

[lentesta]

I’m in the process of trying to determine something related to this. My objective is to complete a WDW46 (http://www.parkeology.com/wdw46-faq) before it becomes WDW46+new rides in such a short time that no one can beat it. Looking for the “optimal path” across 4 parks has been a bit headache inducing.

I actually am using excel in this endeavor! I can’t even guess at how many crowd calendars, wait time charts, and touring plan type things I’ve used at this point. So far just my “final calculations” workbook is at 17 sheets. I’ve picked my optimum dates for the attempt and have planned out 3 tentative schedules. Working on my own time trials is my next step for refinement. (As a single person with FPs and good timing I can now clear Studios in 2 ½ hours.)

. . . however I’ve been trying to make the WDW46 happen since February and I’ve recalculated my plans nearly 2 dozen times. I feel there is no surefire way to know in advance the most expedient path – just a really well educated idea of one.

(attached screen caps of my insanity for hilarity)

This should be doable with the software. I'd have a car waiting on standby too. But I think it's doable.

 

[KeithVH]

Appreciate the discussion everyone. Thank yu for the input. My DW went all cross-eyed when I mentioned my post - she considers this overthinking by a factor of gazillion.

Was aware of TouringPlans but . . . my day job is predictive analytics and, as probably won't surprise anyone near the field, prefer (a) my own algorithms and (b) to do it myself {how any devs do you know blindly accept another's code?}. Will have to go look in more detail at Len's site.

I still think there's room for weather considerations in crowd impact and using some external scoring factor to weigh outcomes to reduce possibilities. Or maybe evaluate/score results with some type of ordinal/linear regression? Things are slow at work, maybe I can get some extra machine time.

 

[englanddg]

Appreciate the discussion everyone. Thank yu for the input. My DW went all cross-eyed when I mentioned my post - she considers this overthinking by a factor of gazillion.

Was aware of TouringPlans but . . . my day job is predictive analytics and, as probably won't surprise anyone near the field, prefer (a) my own algorithms and (b) to do it myself {how any devs do you know blindly accept another's code?}. Will have to go look in more detail at Len's site.

I still think there's room for weather considerations in crowd impact and using some external scoring factor to weigh outcomes to reduce possibilities. Or maybe evaluate/score results with some type of ordinal/linear regression? Things are slow at work, maybe I can get some extra machine time.

I'm sure you and @lentesta could have some interesting conversations. It sounds like a fun project! Your weather considerations add is interesting. I'm fairly sure TouringPlans doesn't take NOAA (or another source) data into account.

That said, the one thing Len has that other sites I've seen do not collect nearly as well is historical data. He can speak for his exact methods, but as I understand it the only organization I can think of that spends more time collecting real historical data is Disney (and they aren't giving that out). And, that is just as important, in many cases more important, as a working algorithm, for accurate analysis.

You can read more about his methods in general here:

http://touringplans.com/walt-disney-world/touring-plans/how-they-work

 

[Tom]

I'm sure you and @lentesta could have some interesting conversations. It sounds like a fun project! Your weather considerations add is interesting. I'm fairly sure TouringPlans doesn't take NOAA (or another source) data into account.

That said, the one thing Len has that other sites I've seen do not collect nearly as well is historical data. He can speak for his exact methods, but as I understand it the only organization I can think of that spends more time collecting real historical data is Disney (and they aren't giving that out). And, that is just as important, in many cases more important, as a working algorithm, for accurate analysis.

You can read more about his methods in general here:

http://touringplans.com/walt-disney-world/touring-plans/how-they-work

Not to mention the LINES app, which allows users to provide real-time, ACTUAL wait times to TP via the app. So, between TP's boots on the ground, and the data being submitted by users, their historical data repository is growing immensely.

 

[englanddg]

Not to mention the LINES app, which allows users to provide real-time, ACTUAL wait times to TP via the app. So, between TP's boots on the ground, and the data being submitted by users, their historical data repository is growing immensely.

As I said, the only data collection mechanism I can think that is better is Disney's current one with MB's and the LE Bluetooth feeding metrics to them. And, there is no way they are handing that data out to anyone. That said, until MDE/Magicbands, Disney's data collection was less accurate than Touringplans, from my experience, strictly due to that. The reason being is they used largely manual methods (such as the "red card on a lanyard") to estimate and record wait times, etc.

MM+ clearly trumps anything that Touringplans could do, but for a third party data collection system, it's about as good as you are gonna find in the Disney Community (imho)...and it's been collecting data for a LONG time compared to MM+.

 

[Tom]

As I said, the only data collection mechanism I can think that is better is Disney's current one with MB's and the LE Bluetooth feeding metrics to them. And, there is no way they are handing that data out to anyone. That said, until MDE/Magicbands, Disney's data collection was less accurate than Touringplans, from my experience, strictly due to that. The reason being is they used largely manual methods (such as the "red card on a lanyard") to estimate and record wait times, etc.

MM+ clearly trumps anything that Touringplans could do, but for a third party data collection system, it's about as good as you are gonna find in the Disney Community (imho)...and it's been collecting data for a LONG time compared to MM+.

On a related note, regarding the "red card" system, if you think about it, if the wait time says 45 minutes on the sign, that's how long the wait was 45 minutes ago. At 2:00pm someone enters the queue with a red card. At 2:45pm they board the ride vehicle. CM scans the card and the Wait Time sign updates (I assume). During those 45 minutes, the line could have doubled in length or reduced to nothing (or anything in between).

I don't know if they'll go to the extreme, but in theory, with the long-range transmitters in use, they could install receivers within queues and track any number of Magic Bands as they move from the beginning to the end of the queue. If they sample enough of them, in shorter time increments, they could produce a fairly accurate real-time "Wait Time".

 

[Daniel Johnson]

As much as I enjoy statistical planing...when in "the world" I just let it all come as it does. We get there early enough and stay late enough...we hardly use fast passes.

 

[englanddg]

On a related note, regarding the "red card" system, if you think about it, if the wait time says 45 minutes on the sign, that's how long the wait was 45 minutes ago. At 2:00pm someone enters the queue with a red card. At 2:45pm they board the ride vehicle. CM scans the card and the Wait Time sign updates (I assume). During those 45 minutes, the line could have doubled in length or reduced to nothing (or anything in between).

I don't know if they'll go to the extreme, but in theory, with the long-range transmitters in use, they could install receivers within queues and track any number of Magic Bands as they move from the beginning to the end of the queue. If they sample enough of them, in shorter time increments, they could produce a fairly accurate real-time "Wait Time".

I'm quite sure that is where it's headed, if it's not already there. Though, I suspect the reporting through the MDE app is rounded to the next 5 minute interval. Again, though, something the Lines app has been doing for a very long time (though with less total accuracy, far better than the red cards.

I can only imagine how much integrity issues they had with the red cards. We were asked to be the "red card" guest several times on our various trips, and my kiddo enjoyed the experience (made her feel important and special), but from a strictly data collection standpoint, it was grossly inaccurate.

The other aspect I'd love to see (and would do if it was my system to design, so I'm sure Disney thought of it) is a real time heat map for the parks. Staggs hinted at it several times in some of his presentations on the subject, so I picture sort of a "mission control" room where they have a heat map up. It's a rather obvious application, but one that I think is cool in concept!

 

[englanddg]

Looking for the “optimal path” across 4 parks has been a bit headache inducing.

Doing all 4 in one day will do that!

 

[lentesta]

Appreciate the discussion everyone. Thank yu for the input. My DW went all cross-eyed when I mentioned my post - she considers this overthinking by a factor of gazillion.

Was aware of TouringPlans but . . . my day job is predictive analytics and, as probably won't surprise anyone near the field, prefer (a) my own algorithms and (b) to do it myself {how any devs do you know blindly accept another's code?}. Will have to go look in more detail at Len's site.

I still think there's room for weather considerations in crowd impact and using some external scoring factor to weigh outcomes to reduce possibilities. Or maybe evaluate/score results with some type of ordinal/linear regression? Things are slow at work, maybe I can get some extra machine time.

We gave up on regression a couple of years ago, moving to gradient boosting machines. It's helped a lot. It also helps to know how much actual waits vary from Disney's posted waits, by ride.

We also take weather in to account, but not yet to update short-range predictions. It would swing the daily predictions a lot, and that tends to upset people. We're considering it for day-of predictions, though.

 

[lentesta]

<snip>

The other aspect I'd love to see (and would do if it was my system to design, so I'm sure Disney thought of it) is a real time heat map for the parks. Staggs hinted at it several times in some of his presentations on the subject, so I picture sort of a "mission control" room where they have a heat map up. It's a rather obvious application, but one that I think is cool in concept!

We have a demo of this. Let me see if I can find it.

 

[HolleBolleGijs]

On a related note, regarding the "red card" system, if you think about it, if the wait time says 45 minutes on the sign, that's how long the wait was 45 minutes ago. At 2:00pm someone enters the queue with a red card. At 2:45pm they board the ride vehicle. CM scans the card and the Wait Time sign updates (I assume). During those 45 minutes, the line could have doubled in length or reduced to nothing (or anything in between).

I don't know if they'll go to the extreme, but in theory, with the long-range transmitters in use, they could install receivers within queues and track any number of Magic Bands as they move from the beginning to the end of the queue. If they sample enough of them, in shorter time increments, they could produce a fairly accurate real-time "Wait Time".

That's always been my problem with the card system. Although I wonder if it's not just based on one card. Like there's probably some sort of average based on all the cards that are out at the moment? For example during MSEP Splash Mountain displays a 5 minute wait. Right after MSEP ends, everyone flocks to the ride, so it won't stay at 5 minutes for long. But sending a card through the line won't give an accurate time because of how suddenly the wait time will jump up. I'm assuming there are other factors that come into play.

In any case, I always find it more reliable to gauge wait times just by looking at where the line is. Of course, that only really works if the queue is outdoors for a significant portion. That's what some other parks do normally (Parc Astérix in Paris comes to mind). They just have signs saying "The line is X minutes from this point," and it's honestly a lot more useful.

 

[Tom]

That's always been my problem with the card system. Although I wonder if it's not just based on one card. Like there's probably some sort of average based on all the cards that are out at the moment? For example during MSEP Splash Mountain displays a 5 minute wait. Right after MSEP ends, everyone flocks to the ride, so it won't stay at 5 minutes for long. But sending a card through the line won't give an accurate time because of how suddenly the wait time will jump up. I'm assuming there are other factors that come into play.

In any case, I always find it more reliable to gauge wait times just by looking at where the line is. Of course, that only really works if the queue is outdoors for a significant portion. That's what some other parks do normally (Parc Astérix in Paris comes to mind). They just have signs saying "The line is X minutes from this point," and it's honestly a lot more useful.

Good point. I suppose, if the entry CM was being diligent about the card distribution (although, likely, they don't know the impact of their actions beyond "these red cards make the wait time change"), they could hand them out every 20-50 guests if an influx of guests came to an attraction that was previously at a low wait.

As the computer sees one card enter and exit quickly, but then sees cards entering at a consistent rate, but exiting at longer and longer intervals, it could hypothesize a new wait time. That's giving Disney a lot of credit though.

As for eyeballing the queue, I've never been successful at that Even with "outdoor" queues, like Jungle Cruise, there are approximately 4.5 miles of queue hidden behind walls and buildings. They can turn a 10 minute queue into a 90 minute queue with the movement of a 3-foot long rope.

 

[HolleBolleGijs]

As for eyeballing the queue, I've never been successful at that Even with "outdoor" queues, like Jungle Cruise, there are approximately 4.5 miles of queue hidden behind walls and buildings. They can turn a 10 minute queue into a 90 minute queue with the movement of a 3-foot long rope.

True. I got very used to it last year with my AP at DLP. I never really timed it, but I just knew whether or not I felt like waiting in the line. Although that usually meant I would skip if it weren't a walk-on haha.