Prashant's Weblog
Random Brain "Droppings" of yet another software engineer

Wednesday, June 28, 2006

Farecast - Good example for Datamining

Last month I along with friends went to Tirumala. As part of the travel plan I took the responsibility to book tickets for the trip. A dolt which I am of first grade, I blindly went to the nearest travel agent and booked the tickets. I paid 800 bucks per person. So when we boarded the bus and were having a casual chat with "neighbours", we were in for a big shock, we had paid at least 150 to 200 bucks more per person. Congratulations. Ok guys now a interesting question;

Q) How to draw cartoon of Prashant?

yeah, I was the bakra. But that gave way to a interesting thought. Is there a way to find the cheapest priced ticket on any given day? Soon we were on the bus, had fun and completely forgot about that idea. But it kindled back to life today when I read about Farecast. They have found a data mining solution to the exact problem I had faced.

Farecast helps you decide if you should buy air ticket now or wait for some more time, so that you get the best deal. They have patented a Data mining algorithm which mines data of close to 50 billion price observations and fits in a predictive model. This model is then used to give suggestions to end used whether they have to buy the ticket or not. The user interface is very intuitive and easy to use. The indicators that hits the user to buy or not to buy the ticket are easy to recognise.

This is really cool stuff. This is perhaps the best example of data mining I have ever come across, probably because after the Tirumala trip I can relate myself with it .

You can know more as to how it works from this page. Also the founder of this start up, Oren Etzioni, formerly Creator of MetaCrawler(I think) has co-authored a white paper of sort called "To Buy or Not to Buy : Mining Airfare Data to minimize Ticket Purchase Price". Really cool stuff. You can get hold of it here. If you want to know more about the startup, you can find it here.


Post a Comment

Links to this post:

Create a Link

<< Home