Is The Connected World Killing Predictive Modeling?

from the the-speed-of-information-is-faster-than-your-algorithm dept

Here's an opinion piece that caught my eye, suggesting that the interconnected world and things like Twitter mean that predictive modeling tools for things like retail sales may be in trouble. Effectively, the argument can be summed up as: information flows faster than your algorithm. I'm not sure it's really necessarily true, but it's certainly worth thinking about. Specifically, he's suggesting that as much as you can try to model "expected behavior" you simply cannot predict or measure sudden flash-fads, which are more common and more impactful with the speed of information these days.

The example he uses is where Michelle Obama recently wore a J. Crew outfit, driving sales of that outfit through the roof. No predictive model could expect that. In some ways, this actually reminds me of another recent story as well. Comedian David Chappelle was apparently in Portland and told a couple people he was planning to show up at a certain city square to do an impromptu midnight outdoor show with a tiny amp and microphone he had just purchased. But in the course of about seven hours, the news spread rapidly via Twitter and Facebook. Chappelle had expected perhaps 200 people, and estimates in the end put the number at many thousands of people (the small amp apparently couldn't amplify his voice enough, and the large crowd became such a concern that he basically sent everyone home after a few attempts to get a bigger amp). While it didn't involve an algorithmic predictive model, this case involved Chappelle's human predictive model that over the course of a few hours you could maybe expect 200 people who would (a) find out about it (b) be nearby and (c) be willing and able to come out at midnight. But the ease of communication changed that equation and made it a lot more difficult to predict the outcome.

Now, of course, the issue with both of these are that they may be outliers. Most other clothing at J. Crew probably followed a typical predictive sales path. And a less famous comedian would probably struggle to get anywhere near 200 people to show up. So, I'd say I'm not at all convinced (as the article posits) that Twitter "confounds" predictive modeling. I think it still requires other random events to occur -- and those have always happened in the past. Perhaps the interconnected nature of the world today can massively amplify a sudden flash fad, but that doesn't necessarily mean you toss predictive modeling out the window. Considering it was such an odd claim to use a few random outliers to damn an entire useful tool, it seemed worth digging a bit deeper... and (surprise!) it turns out that the guy who wrote the article runs a company selling a tool that competes with predictive modeling software focusing on "behavioral analytics." So, his argument is basically a strawman against predictive modeling that uses some outlier data without any evidence of a real trend. I'm certainly interested in the ability of social media to amplify a fad, but I don't think it has a really serious impact on most predictive modeling.


Reader Comments (rss)

(Flattened / Threaded)

  1.  
    icon
    DJ (profile), Jul 20th, 2009 @ 6:03pm

    Kudos

    Well, you still have to give the guy kudos for proactively trying to make the general public dislike his competition. At least he's not suing anyone (yet).

     

    reply to this | link to this | view in thread ]

  2.  
    identicon
    S.P. Gass, Jul 20th, 2009 @ 6:16pm

    Good analysis. Amplified fads or not, companies still need to forecast and schedule employees to handle expected volumes.

     

    reply to this | link to this | view in thread ]

  3.  
    identicon
    NullOp, Jul 20th, 2009 @ 6:42pm

    Modeling

    Sounds to me like we are in the process of writing the models that include Twitter and the like. As for M. Obama, that sort of thing has been going on quite a while. It sounds to me like Twitter, et al simple accelerate the velocity. In short, word spreads faster than expected. At least for now....

     

    reply to this | link to this | view in thread ]

  4.  
    identicon
    Anonymous Coward, Jul 20th, 2009 @ 6:52pm

    The problem is seperating standard events (sale at J Crew) with a celeb sighting / gathering (Hayden Panieterre is naked at the beach right now). The speed that a celeb thing moves has very little to do with how the public perceives and reacts to other things.

    We communicate more quickly, and that will always have an effect. But at what point does a human speed transaction change for that communication? Hmmm.

     

    reply to this | link to this | view in thread ]

  5.  
    icon
    JackOfShadows (profile), Jul 20th, 2009 @ 6:53pm

    Yawn

    Thirty-five years of doing exactly this and it's nice to see someone else waking up to the fact that you do have to put in place the ability to rapidly detect and respond to outlier behaviors. Where he falls short, IMNSHO, is in the anticipated detection and response time. Twenty-four hours, in more than a few cases (epidemiology or financial markets anyone?), is insufficient. Toss in lead times required with JIT manufacturing (J. Crew), and this isn't fast enough by far.

    Actually, it would be kind of cute to incorporate Social Web content into the data warehouse for social trend analysis (and I'm fiddling with this right now: ECMI) but that would almost certainly get stomped on by everyone (EFF, ACLU, ...). I've done autonomous models of this type before. In any case, the standard predictive model is so well entrenched that it will take quite a while, or a few extremely successful players, to change the playing field. The compute cloud may also change this, especially for smaller players.

     

    reply to this | link to this | view in thread ]

  6.  
    icon
    Christopher Gizzi (profile), Jul 20th, 2009 @ 7:45pm

    The models are wrong.

    I think its safe to say that our behavior can always be modeled and predicted. But good measurements come first. It just might be harder for researchers to measure Twitter's historical data.

     

    reply to this | link to this | view in thread ]

  7.  
    identicon
    Anonymous Coward, Jul 20th, 2009 @ 7:47pm

    "behavioral analytics"

    Does behavior analysis fall under the realm of predictive models? I mean, what's the purpose of it if it can't make predictions? What good would his behavioral analysis be if it can't make predictions? Seems rather ironic that he's saying predictive modeling doesn't work yet he's trying to sell something to help people make predictions.

     

    reply to this | link to this | view in thread ]

  8.  
    identicon
    Lawrence D'Oliveiro, Jul 20th, 2009 @ 8:07pm

    Fundamental Theorem

    Freer information flow is supposed to lead to more efficient markets. But all it really does is increase the instabilities.

     

    reply to this | link to this | view in thread ]

  9.  
    icon
    scarr (profile), Jul 20th, 2009 @ 8:22pm

    Re: The models are wrong.

    An an (ex?) engineer, I have to agree that it's a matter of having the correct measuring devices in place. You still need a model to account for them, but ignoring pieces of the puzzle will always corrupt the image you're trying to create.

     

    reply to this | link to this | view in thread ]

  10.  
    identicon
    Anonymous Coward, Jul 20th, 2009 @ 10:25pm

    Re: Fundamental Theorem

    So why bother posting on techdirt unless you are purposely making the market less efficient and increasing the instabilities. Stop posting if you think that it only makes things worse. And for that matter cancel your internet connection, get rid of your television(s) and read the newspaper 1 week late. That way you can practice what you preach.

     

    reply to this | link to this | view in thread ]

  11.  
    identicon
    Jack Pine, Jul 21st, 2009 @ 2:01am

    Re:Fundamental Theorem

    Under rational expectations, instability should disappear, unless there is a singularity. The latter would appear if non-equilibrium outcomes persist due to decision makers being led astray by outliers - something greater information would preclude, not enhance. While there may be a tatonement driven adjustment process, so long as it is convergent (assured by non-existence of singularities, and rational behavior) stable equilibrium is possible. It is not more information but human behavior that leads to volatility! More information results in greater learning.

     

    reply to this | link to this | view in thread ]

  12.  
    identicon
    Lawrence D'Oliveiro, Jul 21st, 2009 @ 4:29am

    Re: Fundamental Theorem

    Jack Pine wrote:

    Under rational expectations, instability should disappear...

    As is well known, it doesn’t.

    Think of the economy as a dynamical system. Or, for a simple (but mathematically useful) analogy, consider the suspension system of a car. The springs are the error-correcting mechanism: they represent how the market adjusts to new information. But there is also the inertia of the car body: this represents expectations that things will continue the way they have in the past. So when you hit a bump, the springs try to move the car body to a new position, but the inertia makes it overshoot. Then the springs pull the other way, and it overshoots again. So it ends up oscillating. If the pattern of bumps matches the resonant frequency of the system, the result can actually damage the car. Not to mention the effect on the occupants...

    Which is where the shock absorbers come in. They are designed to dissipate energy from the system, and slow down the rate at which it responds to stimuli. In short, they waste energy, and by doing so, they dampen down the tendency towards instability. The equivalent in the economy is Government regulation. That, too, reduces the efficiency with which the system responds to new stimuli. And by doing so, it makes it less likely to self-destruct.

     

    reply to this | link to this | view in thread ]

  13.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 6:03am

    Re: Re: Fundamental Theorem

    Actually I would argue that government regulation has contributed to the instability. For instance the government started granting cable companies monopolies and what ended up happening? It made it easier for special interest groups to censor important news that people wanted to hear, cable companies started offering trash shows that cost them less and they started charging more for their service, and the cable companies started putting more and more commercials on television. Not to mention the effort they put into suppressing innovation (ie: DVR's like Tivo though the courts ruled against it. Why innovate when you already have a monopoly, it doesn't help you gain market share. Fore example, when more competition was introduced into cell phones and cell phone companies then cell phone companies started to innovate by offering phones with word processors, text messaging plans, the Internet, prices started going down, etc... I remember before that a 1/4 mile walkie talkie would cost $100 but then after competition was introduced in the cell phone industry no one wanted walkie talkies anymore because they had cell phones. Now you can get a 5 mile walkie talkie for like $30). So what do people do? They look for alternatives. Likewise, if you start regulating the Internet in favor of rich and powerful corporations at the expense of everyone else (like they did with cable) people will naturally look for alternatives. And when competition was introduced to Lan Line companies a few years back is when Caller ID, 3 way calling, and all sorts of other innovations started to become available. Now that there is once again a monopoly you see no new innovations in Lan lines. Why innovate when you own the market?

     

    reply to this | link to this | view in thread ]

  14.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 6:08am

    Re: Re: Fundamental Theorem

    "That, too, reduces the efficiency with which the system responds to new stimuli."

    I would argue that a system that quickly responds to new stimuli is an economically more efficient system. If someone is going to punch you and you respond 1 minute late you're already hurt by then. Allow people to respond to stimuli, lets not have the government force us not to be informed about important news that we should know. This sounds like an attempt to limit our freedom of speech in ways that the government has no business doing. After all one of the assumptions about free markets and why they cause more efficiency is that information is freely available.

     

    reply to this | link to this | view in thread ]

  15.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 6:29am

    Re: Re: Fundamental Theorem

    The rich and powerful entities just want things to be stable in a way that maintains the status quo. Cable companies with government granted monopolies don't want competition introduced because they want laws that benefit them and no one else. If they have to compete that means they actually have to work for their money instead of extort it out of customers. So yeah, stability is great for the status quo but it shoulnd't be the governments job to maintain the status quo and ensure that the rich and powerful can maintain their status and not work for it (ie: by not innovating and by producing less product at a higher price, etc...). They should have to compete so that aggregate output increases. The whole purpose of economics is to increase aggregate output, not to have the government keep stability in a way that benefits the rich and powerful at the expense of everyone else (by maintaining the status quo).

     

    reply to this | link to this | view in thread ]

  16.  
    identicon
    Bradley Stewart, Jul 21st, 2009 @ 7:09am

    Predictive Modeling For Me When I Worked

    was always something that I did on the back of an envelope and I was pretty good at it. I once set up a company's whole Holiday Season ordering in six and a half hours in a bar with nothing but a few stacks of catalogs and a calculator. It worked out almost prefectly. This was before the age of computers that did things like Predictive Modeling. I wish that I had a tool like this program at the time. I am sure that It would have helped me. Now that I'm a retired old codger I will only pass along this thought to any younger folks that use a tool like this for what ever purpose. The person or persons that designed and or set up this program may have done it in a bar and spent a lot longer than the six and a half hours that I spent there.

     

    reply to this | link to this | view in thread ]

  17.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 7:24am

    Re: Re: Fundamental Theorem

    If someone in the free market finds a better way to serve the market such that people switch to that entity then that's good. The free market allows for whoever can best serve the market to do so. Sure, this may cause instability if one entity with dominant market share suddenly does a poor job at servicing the market and some other entity finds a better way to server the market such that everyone switches to that new entity. But so what? That's market efficiency, it's not the governments job to maintain the status quo by keeping the entity that does a worse job at serving the market at the expense of the person who can better server the market. This is bad for society, we should allow those that can best server the market to do so.

     

    reply to this | link to this | view in thread ]

  18.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 7:25am

    Re: Re: Re: Fundamental Theorem

    sp/server/serve

     

    reply to this | link to this | view in thread ]

  19.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 7:40am

    Re: Re: Re: Fundamental Theorem

    and sure, faster communication (ie: the Internet) could mean that people will find out who can best serve the market faster which may lead to instability in a way that's bad for those who do a poorer job at serving the market. So what?

     

    reply to this | link to this | view in thread ]

  20.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 8:04am

    Re: Re: Fundamental Theorem

    Also note, the average person acts in his/her own best interest and when receiving new information they tend to act in their best interest based on that information. What you're asking for are laws that make it more difficult for the average person to act in his/her own best interest. Who's interest does this serve? It serves the interests of the status quo, and more specifically, the rich and the powerful at the expense of everyone else.

     

    reply to this | link to this | view in thread ]

  21.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 8:45am

    Re: Re: Fundamental Theorem

    and if the government takes away the means for us to act in our own best interest (ie: faster communication) then who is expected to act in our best interest? The government? I think not. Big corporations? I find that hard to believe. What you are asking for is a state of tyranny where some third party is expected to act in our own best interest instead of allowing each of us to act in our own best interest. We don't know that a third party (ie: the government or corporations) intends to act in our own best interest but we know that we intend to act in our own best interest.

     

    reply to this | link to this | view in thread ]

  22.  
    identicon
    Anonymous Coward, Jul 21st, 2009 @ 10:15am

    The only thing that will ruin predictive modeling is knowing the future.

    Predictive modeling just now has to take into account how fast information flows. This is just an evolution.

     

    reply to this | link to this | view in thread ]


Add Your Comment

Have a Techdirt Account? Sign in now. Want one? Register here
Get Techdirt’s Daily Email
Save me a cookie
  • Note: A CRLF will be replaced by a break tag (<br>), all other allowable HTML will remain intact
  • Allowed HTML Tags: <b> <i> <a> <em> <br> <strong> <blockquote> <hr> <tt>
Follow Techdirt
A word from our sponsors...
Essential Reading
Techdirt Reading List
Techdirt Insider Chat
A word from our sponsors...
Recent Stories
A word from our sponsors...

Close

Email This