Tuesday, July 10, 2012

Data Analytics: My two Cents



I was going through an article from IBM press release the other day.  The title read “Charleston Police Department Aims to Reduce Crime Using IBM Predictive Analytics” [1]. Three things which came up to my mind at that point were,
                             
  1. The movie “Money ball” –The good
  2. Crash of AF 447 – The bad
  3. The movie “Minority report” – The quintessential question


Money ball[2] movie is the poster boy of “Data Analytics”. This movie is based on a true story where Oakland Athletics general manager Billy Beane (Brad Pitt) takes help of Sabermetrics[3]  (read “Data  Analytics”) which helps his team winning 20 consecutive games, an American League record.  Gee! Which gives an impression that, why can’t this strategy be implemented by all team coaches?  Wait a minute! there may be a catch, lets park this question for now.

On 9th June 2009 an Air France passenger jet crashed into Atlantic ocean, killing all 216 passengers and 12 crew. One of the reasons for the crash is stated to be faulty instrument reading. One thing that caught my attention is,

[4] “Despite the fact that they were aware that altitude was declining rapidly, the pilots were unable to determine which instruments to trust: it may have appeared to them that all values were incoherent”

     “Data Analytics” is analogous to instruments which help us in our decision making. Personally, I feel Analytics are spit out using algorithms which use some correlations, most of the time. Now comes the red herring, you are responsible for, how much to depend on this data and how to use the data.  This answers our parked question.

     In the movie “The Minority Report”, Spielberg –The director- tries to address a quintessential question of   “free will vs. determinism”.  To put it in plain English, are all the actions we intend to perform, pre-determined or is there anything like "free will". Also, if and when, ‘data analytics’ attains the ‘holy grail’ of predicting the future, can the State use such data to punish the would-be guilty before they actually commit crime and other things around it.

     Personally, I would put my buck in  “Data Analytics”  where the end result is not used to decide on life and death, in anyway. If it’s related to life and death, first I would rely on instincts, judgment and training later, think about what “Analytics”say.

    It sounds logical to me, for a cop to wear his bullet proof vest and residents to lock their homes, even when the neighborhood is “statistically” stated as zero crime area for past several years.



 This is written for my office magazine. Thought to publish on my personal blog




Sources:
1.       Charleston Police Department Aims to Reduce Crime Using IBM Predictive Analytics
http://www-03.ibm.com/press/us/en/pressrelease/37985.wss
2.       Moneyball (film)
http://en.wikipedia.org/wiki/Moneyball_(film)
3.       Sabermetrics
http://en.wikipedia.org/wiki/Sabermetrics
4.       Air France Flight 447
http://en.wikipedia.org/wiki/Air_France_Flight_447









1 comment:

Unknown said...

I do agree that when its the question of life and death we need to follow our instincts... but its more likely that you will be alive only if you are trained for such scenarios. And in your example, the AF flight pilots did not have enough training for those altitudes over the Atlantic ocean. That area is said to be a special climatic zone. Good article Kalyan.