Books on machine learning start with some grandiose reason why machine learning is awesome, and a simple example and current application is "basket analysis". Basket analysis refers to computation done on the stuff you buy at any grocery store. This can be done without a loyalty card, but it makes it easier to see more correlated trends if you have access to particular customers purchasing trends.
To make the long story short, the reason for basket analysis follows a logic like this: "If customer A purchases product X and Y often in the same shopping trip, placing these products closer together allows for cross-selling and increase revenue."
This sounds good but you have to be careful about how you go about doing this:
If I am looking to purchase product Y (not starting with X first), and I go to where I remember seeing this product, or where it makes sense for it to be stored, and I don't find it there, I will get frustrated. So, one has to put product Y in two places. Product Y next to X can triger reminders to the casual customer, so there's value there, however, if you are making it harder for your regular customers to find what they're looking for, than that's not good.
Being vegetarian, I was quite frustrated when Kroger moved their meats near the meat substitutes because they stored beer there or for some other weird reason. That made sense for some customers, but frustrated me.
Firebug
MeasureIt
ColorZilla
Delicious Bookmarks
Html Validator