Spirited News, Observations & Thoughts IV

[PhotoDave219]

This has been prevalent in many industries. The problem is that it leads to data glut and completely incorrect or inconsequential conclusions. I see it all the time in my field, which is scientific research. We have all this fancy equipment nowadays that spits out info at the press of a button. Scientists go in and do these incredibly sophisticated analyses, come out with reams of data.

Me: Why did you do this test?
Them: Because um....

They then proceed to form a conclusion based on the data they have just gotten. Then work backwards and claim this is what they were looking for all along. And you see the problem...instead of an experiment guided by a fundamental question, coming up with a hypothesis and testing their hypothesis, they randomly accumulate all this data, and then try to back engineer a justification for doing it, often by placing emphasis on all the wrong things.

I suspect this Data Mining project at Disney is the same thing. Instead of starting at a fundamental problem that needed to be solved, Disney looked at their data and started a project to address a nonexistent problem.

Ok so you can do an ANOVA test. Why did you do an ANOVA test? What does the average represent? What does the variance represent? Why does it matter that there's a variance in averages?

I want to believe that the behind the scenes tracking is merely industrial engineering and measuring the dynamics of guest behaviors in a theme park environment. Really studying exactly what guests really do.

I want to believe that but in the nine years I've been around this company and seen the way it operates, in that how it markets things and sells them, it makes me think that they intend to use all this information for ways to further sell things to you and possibly sell your data to third parties. I have no specific information that concrete says this, its just what my gut says.... plus the hints of things others have said. (Palantir? I'm curious)

 

[Cesar R M]

I want to believe that the behind the scenes tracking is merely industrial engineering and measuring the dynamics of guest behaviors in a theme park environment. Really studying exactly what guests really do.

I want to believe that but in the nine years I've been around this company and seen the way it operates, in that how it markets things and sells them, it makes me think that they intend to use all this information for ways to further sell things to you and possibly sell your data to third parties. I have no specific information that concrete says this, its just what my gut says.... plus the hints of things others have said. (Palantir? I'm curious)

I wouldnt be surprised.
Most probably use these to bombard you with offerds suited to what you do, a la Amazon.

 

[dupac]

And there is part of their problem, ANOVA is analysys of variance, More years ago than I care to admit was running MANOVA which is MULTIPLE analysis of variance ie looking for WHICH is the determining factor from a variety of factors much more complex and powerful technique.

You could always do a Tukey test... lol

 

[dupac]

This has been prevalent in many industries. The problem is that it leads to data glut and completely incorrect or inconsequential conclusions. I see it all the time in my field, which is scientific research. We have all this fancy equipment nowadays that spits out info at the press of a button. Scientists go in and do these incredibly sophisticated analyses, come out with reams of data.

Me: Why did you do this test?
Them: Because um....

They then proceed to form a conclusion based on the data they have just gotten. Then work backwards and claim this is what they were looking for all along. And you see the problem...instead of an experiment guided by a fundamental question, coming up with a hypothesis and testing their hypothesis, they randomly accumulate all this data, and then try to back engineer a justification for doing it, often by placing emphasis on all the wrong things.

I suspect this Data Mining project at Disney is the same thing. Instead of starting at a fundamental problem that needed to be solved, Disney looked at their data and started a project to address a nonexistent problem.

Ok so you can do an ANOVA test. Why did you do an ANOVA test? What does the average represent? What does the variance represent? Why does it matter that there's a variance in averages?

I am currently wrapping up my last 9 hours of school (about time) and sitting in a senior design presentation, one of my classmates tried to use MRP (material requirements planning) to prove something about production capacity or throughput or something. He got raked over the coals by our productions and operations professor. "So you can do an MRP. But why did you do it? I don't understand what you're trying to support"

It's one thing to have the tools; it's another to know how to use them.

 

[Nubs70]

I want to believe that the behind the scenes tracking is merely industrial engineering and measuring the dynamics of guest behaviors in a theme park environment. Really studying exactly what guests really do.

But does another study of customer behavior and the sale of its findings provide sufficient ROI to justify the level of investment? (Palantir, I'm curious)

 

[PhotoDave219]

But does another study of customer behavior and the sale of its findings provide sufficient ROI to justify the level of investment? (Palantir, I'm curious)

I really want verification that Palantir is actually involved (google that, it was in the reading if you missed it a few months back) and if they actually are, to what level, etc.

And frankly this is the question the Board of Directors needs to be asking. Where is our corporate governance?

 

[GoofGoof]

But does another study of customer behavior and the sale of its findings provide sufficient ROI to justify the level of investment? (Palantir, I'm curious)

When they spread it over 6 parks it should ease the pain and spread the costs a little. Sorry DL fans, but it's coming. It's only a matter of time.

I think the system is far more advanced than a simple study of customer behavior. They are attempting to predict your individual actions based on your personal trends and habits. Most similar studies use very general demographic data to predict general trends. The scale and scope of the system has never been tried before. The only thing that probably comes close is Amazon and they obviously figured out a way to make a lot of money so who knows, maybe the system really will earn an ROI that meets or exceeds expectations. None of that really matters to me as a guest, but it's a fascinating case study for some business school.

 

[ford91exploder]

I really want verification that Palantir is actually involved (google that, it was in the reading if you missed it a few months back) and if they actually are, to what level, etc.

And frankly this is the question the Board of Directors needs to be asking. Where is our corporate governance?

IF Palantir IS involved it would be incredibly disturbing, - Readers Digest version of posts Palantir is a software company funded by the CIA for the purpose of developing and purchasing software for use by the intelligence community. Even the name from LOTR is descriptive the Palantir's were seven stones which were capable of 'seeing things the Enemy wanted to keep hidden'.

 

[Darth Sidious]

You really should confine your comments to something you understand. Canada is not a colony, not even a nominal one. Those shackles were thrown off long ago.

Actually, I think he's referring to how technically Canada is a 'constitutional monarchy' where the queen is a figure head of sorts.

http://www.parl.gc.ca/About/Senate/Monarchy/SenMonarchy_00-e.htm

 

[ford91exploder]

This has been prevalent in many industries. The problem is that it leads to data glut and completely incorrect or inconsequential conclusions. I see it all the time in my field, which is scientific research. We have all this fancy equipment nowadays that spits out info at the press of a button. Scientists go in and do these incredibly sophisticated analyses, come out with reams of data.

Me: Why did you do this test?
Them: Because um....

They then proceed to form a conclusion based on the data they have just gotten. Then work backwards and claim this is what they were looking for all along. And you see the problem...instead of an experiment guided by a fundamental question, coming up with a hypothesis and testing their hypothesis, they randomly accumulate all this data, and then try to back engineer a justification for doing it, often by placing emphasis on all the wrong things.

I suspect this Data Mining project at Disney is the same thing. Instead of starting at a fundamental problem that needed to be solved, Disney looked at their data and started a project to address a nonexistent problem.

Ok so you can do an ANOVA test. Why did you do an ANOVA test? What does the average represent? What does the variance represent? Why does it matter that there's a variance in averages?

Exactly - Scientists are bad about this but executives are even worse especially since most of them got a D- on 'Introduction to College Algebra' so they see these tools and misuse them leading to the 'Garbage IN, Gospel OUT' syndrome. Most executives would be better of sticking to 'CROSSTABS' those who labored with SPSS and it's derivitives know exactly of what I speak.

I use these tools for analyzing performance data of network components and systems so I have a clear idea of what I am analyzing and why.

P.S.
'
To the Context impaired the standard usage for GIGO is 'Garbage In Garbage Out', There is another use however for people who use 'Big Data' without understanding the tools and that is 'Garbage IN, Gospel OUT'

Executives who misuse big data say "The Computer is Never Wrong", Yes it is and frequently, while the computations are numerically correct the model used is invalid and therefore so are the results.

 

[Nubs70]

Exactly - Scientists are bad about this but executives are even worse especially since most of them got a D- on 'Introduction to College Algebra' so they see these tools and misuse them leading to the 'Garbage IN, Gospel OUT' syndrome. Most executives would be better of sticking to 'CROSSTABS' those who labored with SPSS and it's derivitives know exactly of what I speak.

I use these tools for analyzing performance data of network components and systems so I have a clear idea of what I am analyzing and why.

Executives fail in that they believe correlation equals causation. In the case of some environmental topics of our day, data that does not support correlation is simply discarded. The remaining data is used to support correlation therefore "proving" causation.

 

[ford91exploder]

Executives fail in that they believe correlation equals causation. In the case of some environmental topics of our day, data that does not support correlation is simply discarded. The remaining data is used to support correlation therefore "proving" causation.

Correlations are interesting but do they really mean anything, Correlations are easy because people think they understand them and remember while executives generally have very high social intelligence they generally have average general intelligence, Tech executives tend to be a bit different in that they generally have significant professional accomplishments in their field of expertise before becoming executives, They also tend to have much lower social intelligence than business executives.

People like this do a disservice to science and management because usually it's the outliers which are more interesting and pursuing them tends to yield groundbreaking results if you have never read 'The Cuckoo's Egg' i would recommend it as it details how a persistent 0.02 cent error in a timesharing account balance led to one of the largest cases of computer espionage in history NASA, Military, Govt and University computers were systematically compromised. and it was only found because some one decided to run down an outlier.

 

[Nubs70]

Correlations simply point to an area for further investigation. I come from a continuous commodity manufacturing background. I use CUSUM inflection point/timeshift analysis combined with correlation to find persistent locations of equipment issues. This method only points to an area for further in-depth investigation. The causal element is most often a mechanical deficiency rather than an operational or control deficiency.

 

[ford91exploder]

Correlations simply point to an area for further investigation. I come from a continuous commodity manufacturing background. I use CUSUM inflection point/timeshift analysis combined with correlation to find persistent locations of equipment issues. This method only points to an t area for further in-depth investigation. The causal element is most often a mechanical deficiency rather than an operational or control deficiency.

Interesting it looks like SPC has gone up several levels since my last visit to a real factory floor. And as a treat I bring you this from XKCD

 

[nytimez]

Executives fail in that they believe correlation equals causation.

This part's correct.

In the case of some environmental topics of our day, data that does not support correlation is simply discarded. The remaining data is used to support correlation therefore "proving" causation.

This is talk-radio paranoia used to justify inaction on the "environmental topics of the day." I could argue this one all day with you, but I'm guessing this isn't the place for it.

 

[PhotoDave219]

IF Palantir IS involved it would be incredibly disturbing, - Readers Digest version of posts Palantir is a software company funded by the CIA for the purpose of developing and purchasing software for use by the intelligence community. Even the name from LOTR is descriptive the Palantir's were seven stones which were capable of 'seeing things the Enemy wanted to keep hidden'.

Thank you for summing that up for the rest of the class. That's much better way of putting it that the way I have a past.

We've heard rumors that they are involved in the backend, That being the analytical databases. Some have claimed to have possession of documents that prove this. I haven't seen them but I believe there's something to this.

 

[The Visionary Soul]

Anyone see the story on the main page about FP+ testing for Swan and Dolphin? According to the update they will begin testing soon and will be the first non-Disney property to use the system. What I found interesting is they are not giving out magic bands or allowing advanced FP+ reservations. Guests of the Swan and Dolphin will only be allowed to use the FP+ kiosks in the park. I know many speculated that FP+ would become a Disney hotel only perk. Is this the first sign of that or is it just related to testing? I guess we'll need to wait until next year to find out if that becomes permanent. The testing goes into January.

This means two things:

Extra Magic Hours will be going away in the evenings eventually permanently.

They will be getting rid of normal paper Fast Passes eventually, everyone will be on the FP+ system.

The thing is, we knew both of those a long time ago.

 

[Nubs70]

This part's correct.



This is talk-radio paranoia used to justify inaction on the "environmental topics of the day." I could argue this one all day with you, but I'm guessing this isn't the place for it.

I do not know how someone can claim a hypothesis is true when they simply remove a portion of a data set that interferes with a true hypothesis. If data is removed, there must be justification.

 

[GoofGoof]

This means two things:

Extra Magic Hours will be going away in the evenings eventually permanently.

They will be getting rid of normal paper Fast Passes eventually, everyone will be on the FP+ system.

The thing is, we knew both of those a long time ago.

The part that was speculation was that FP+ was going to be for resort guests only. Disney came out and said it would be available to all guests. The loophole is that off property guests won't be able to book in advance. I know this was speculate about a lot but this seems to be the first indication that they are actually going through with the plan. That's the part I find interesting.

 

[nytimez]

I do not know how someone can claim a hypothesis is true when they simply remove a portion of a data set that interferes with a true hypothesis. If data is removed, there must be justification.

Arguing climate is like arguing evolution; people who don't believe it, won't believe it - no matter what. But, again, this ain't the time or place for that.