In a few of my classes, the topic of data mining came up.
Basically, it comes down to taking existing data and drawing conclusions from it. This can be seen, for example, if you have an amazon.com account. Amazon shows you suggestions of other items that you might be interested based off of what you are viewing. The way that is done is through a process called data mining.
Basically, it doesn't take a genius to say if someone buys season 1 of a TV show (say, House, M.D.), they may be interested in season 2 of that same TV show. However, if you like House, M.D., you might also be interested in something that may seen completely different - let's say, a Sherlock Holmes mystery. Amazon can tell that by taking everyone's past purchases and analyzing what people who bought a certain DVD liked, and comparing that to your past.
There are many other places where this happens - for example, credit card companies use data mining to guess if a transaction is fraudulent or not - based on your purchasing habits. For example, my credit card was rejected when I tried to go skydiving, because based on what I've charged, they thought that it was so out of character for me that they blocked the transaction, and I had to call in and verify that it was indeed me.
Another place that this happens is on sites like Netflix and Blockbuster Online. Of course, they want you to enjoy movies, but the underlying reason is so that you come back and stay a subscriber. Therefore, it is best for them to have a good data mining algorithm in place to show you movies (and therefore keep you subscribing).
As a matter of fact, if you can provide them a better algorithm than they have, you can be eligible for prize money - up to a million dollars, if your data matching algorithm is substantially better than theirs.
Web Link:
Thanks to Brian E. for the heads up!
Computing is such a large topic, that no matter how good you are, there is always something to learn. I personally feel like I learn something new every day. Knowledge is power, and knowledge should be free.
Showing posts with label netflix. Show all posts
Showing posts with label netflix. Show all posts
Saturday, December 13, 2008
Friday, August 15, 2008
Netflix Down
Netflix (one of the two companies that provide DVD rentals through the mail, along with Blockbuster), has been hit by huge technical problems.
If it were only one site (they have 55 shipping centers), it would be one thing - sometimes problems happen locally. However, for it to happen company wide indicates some sort of failure somewhere.
Rumors suggest that the problem may have been caused by database administrators updating a program called Oracle (a very large database management software package) and blowing things up. However, there's always the possibility that they were hacked too. We'll see.
The service has apparently been down since Tuesday and only a few movies have managed to be shipped in the last three days.
For the record, I haven't heard anyone suggest hacking, but I always have this idea in the back of my mind when I hear stories like this.
For the record, I've used Netflix in the past and was always pretty happy with their service.
Link to article
Subscribe to:
Posts (Atom)