Realities of Predictive Analysis

  • Comments posted to this topic are about the item Realities of Predictive Analysis

  • I'm happy being a skeptic.

    When I'm right, I'm happy.

    When I'm wrong, I'm still happy.

    Disaster Recovery = Backup ( Backup ( Your Backup ) )

  • I used to work for one of the big 3 advertising agencies.  A typical response rate to a direct mail campaign varied around typical figures

    • To consumers, around 5%
    • To businesses, around 2%

    That tells you that in the vast majority of cases we weren't that great at predicting response rates, merely better than random chance.

    Our budget for acquiring a list of people to mail and the material we mailed out was based on those response rates.  An uplift of even 1% could make the campaign highly profitable.

    One of the challenges we faced was in those list of addresses where we picked the most probable responders.  If we picked the 10% most likely to respond, and got a better than average response rate, the client would want to mail more of that list.  That 2nd mailing would always disappoint.  The idea that we'd already skimmed the cream off the list was one the clients seemed to have trouble grasping.

    Preparing data and helping with analysis was what got me into databases in the 1st place.  I still find it a fascinating subject though the discipline has outgrown me.

  • Understanding this concept is critical in poker. Just because the odds are in your favor does not mean you'll win every time.

    Dealt AA? Great, that's the best starting hand! Against a single player you'll lose 20% of the time!

    You can always tell a good player from a bad one -- the bad ones only see "I'm a favorite to win" and extrapolate to "I will win every time."

Viewing 4 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic. Login to reply