Friday, September 21, 2007

To be a Data Miner

I often mention in free talks with my friends if I can make my clock back to 13~15 years ago (high school), I take 'design' as my major, and if I can go back to 10 years ago, I take several courses in biology (life science) as a secondary major.

I spent 10+ years in industrial engineering (now, became industrial and management engineering in my university without any changes in course works about management) - BS, MS, PhD, and PostDoc also. It is not a secret my recent concern is data mining. So, I thought what is the best way to be a data miner. A critical condition is to be teen age if you are not so active and smart as your high school age. If you are still hungry about new knowledge and eager to spend another years for that new things, that is not critical.

First, take mathematics as your primary major in university.
Second, take biology as your secondary major in university. Chemistry is also good.
If you have extra time at those periods, taking several computer science courses such as data structure, algorithm, programming, etc is very useful.
Third, take computer science as your MS program, especially take course works on or labs in AI, machine learning, pattern recognition, NLP, etc.
Fourth, take electronic engineering as your PhD program, especially labs in signal processing or image processing.
This is the full course to be a data miner having both theoretical knowledge and application backgrounds. Mathematics gives you probability and statistics as well as other useful things, CS gives you programmability and AI, and also some text/web mining (i.e., NLP), biology, chemistry, and EE give you the applications on bioinformatics, cheminformatics, or signal/image processing.
An important thing is I'm not sure the next 10+ years. At this time, data mining or AI looks good. Design is also good, but everything has its own life cycle - birth grown-up death, if possible recycling.

I'm mad to be an IEr, but I can design this integrated scheme because I'm an IEr.