1 Replies - 552 Views - Last Post: 26 April 2013 - 08:00 AM Rate Topic: -----

#1 modi123_1   User is online

  • Suitor #2
  • member icon

Reputation: 14045
  • View blog
  • Posts: 56,205
  • Joined: 12-June 08

Data Mining - worth the time?

Posted 26 April 2013 - 07:43 AM

I have some free time at work and I have been poking at our giant glob of data wondering if, with the right stick (or downward angle) I can elicit some interesting results. I know enough about data mining to explain (using hand waving) concepts and sort of push it into a black box on my state diagram, but I am looking for more of an end game pay off discussion.

Joints like walmart, google, The Feds, etc mine the hell out of our information... making odd connections, but as smaller business users has anyone found it useful? Outside of going "huh - that's a pretty strange coincidence that ten percent of our clients avoid booking on the full moon, but all of those ten percent have green eyes, and they have left handed tendencies when traveling east south of the Mason-Dixon line.". You know - odd esoteric baseball trivia that may or may not help out the business cause/bottom line.

So, yeah.. does anyone do 'data mining' as part of their job? Have you seen results? Is this more of a "you need to have a good question going into a well setup model" versus "throw data at the wall and see what sticks"? Were there any super cool "a-ha!" moments that data mining provided? How about bad connections you shouldn't have acted on? Have you tried and just gave up after some sort of blocking issue?

If it makes a world a difference - no, I am not using Oracle.. MSSQL 2012 here... general .NET usage that comes with it.

Resources I am looking through:


Is This A Good Question/Topic? 0
  • +

Replies To: Data Mining - worth the time?

#2 jon.kiparsky   User is offline

  • Beginner
  • member icon

Reputation: 11041
  • View blog
  • Posts: 18,857
  • Joined: 19-March 11

Re: Data Mining - worth the time?

Posted 26 April 2013 - 08:00 AM

I sit by the data warehouse team, so I get to talk to them about what they're working on. They do a certain degree of investigation into the data, but none of it is very systematic. One ongoing project that I like is trying to examine expected versus trade volume for a given day. "Expected" depends on many factors - time of the month, day of the week, holidays, time of year - so there's a lot of different curves to think about.

Mostly, it's a seat-of-the-pants effort. "I wonder what happens if we try this factor and this one..."

The end goal is nicely ambitious: they want to be able to spot issues with the company's software by anomalies in trading volume. I'd really like to see solid results from it, but so far it's still mostly a project for slow afternoons.
Was This Post Helpful? 0
  • +
  • -

Page 1 of 1