For the record

They first gave us a topic to write on, 'Recession is the mother of innovation.' I highlighted the opinion that innovation will come with governmental spending on newer sectors, jobs that have been lost were mostly to cater to an artificial demand which is no more. Innovation in newer sectors to broaden the base of the economy is the only way to jump-start the system back to life again.

Now, the interview:

Panel Desc: two senior people, one of whom had been on my panel last year too , and had done me in with an acads question... I was thoroughly amused by the prospect of facing the same panel again. Very nice, and very senior people.

P1: Come in, CodesmithCipher.
Cipher:
P1: Hoye! Whoever taught you that. Take your seat young man, you don't need to wait till I ask you to take a seat. 
(No point taking chances people, had I sat down he might have wanted to take my case on being stupid enuf not to ask)
Cipher: Big shameless grin, no sir no one taught me that, I'll take the seat. Another big grin.
P1: So, you're with , tell us what you do, I mean, what is your role out there, what do you do... 
Cipher: Err, umm, (clumsily)we work in the domain of knowledge discovery... we try and identify risks and opportunities for a client by creating insights in transactional data.
P1: wow. Knowledge discovery, risks, opportunities, Can you simplify that for us :P
Cipher: We work on supervised learning methods to identify the dependence of response variables on other, given data.
P1: So that is like, ANN's and all (artificial neural networks: a machine learning method)?
Cipher: no sir, we dont use ANN's and other machine learning tools... we stick to statistical methods mostly.
P1: Like what all?
Cipher: We do stuff like regressions, logistic, linear, decision trees etc.
P1: Hmm Hmm Hmm regressions eh? (Broad grin). 
Cipher: Sir, yes sir.
P1: So do you know what CART is, Cipher? I mean, you must have used it, right?
Cipher: Yes sir I do...
P1: So can you tell me what CART is, what Logistic Regression is, how are they different, when do you use CART, when do you use Logistic Regression, and when would you want to use a combination of the two, and how?

Cipher: Certainly sir... Cart is about...
P1 Before that, can you tell me what does CART stand for???
Cipher:  Sir Cart... err, umm... It stands for err... (and then it struck me) Sir it stands for Classification and Regression Trees.
P1: Yes, thankyou. Continue now.
P1: So sir, CART is essentially used, in context of the work we do, to identify sub populations from a large sample size that are markedly different from one another as regards the behaviour of the response variable. Like, in case of a categorical variable, maybe a highly different response rate, or in case of a continuous variable, a large difference in avg. (Mostly on track, but with great deal of clumsiness).

P2: Joins in: Okay, so Cipher do you remember any bit of electrical engineering from college?
P1: How wouldn't he know? Just been seven months out of college. He had better known!
Cipher: Broad 32-all-out grin.
P2: So Cipher, what is a switchgear?

Cipher: Sir, so its about... 
P2: Why are you smiling? And on this particular question?
Cipher: Sir, I had taken IIMA interview last year too, and you asked me the same question 
P1: So, you know what a switchgear is now?
Cipher: Yes sir, I went back last year, and read about it. Didn't answer it then. 
P1: Hmmm! So Cipher lets try and calculate the probability of occurence of such an event. I mean, can you find me the probability that you will encounter the same panelist for two consecutive years?
Cipher: Certainly sir. We will need to take a few numbers as assumptions... 
P1: yea yea sure
Cipher: Lets assume the pool of profs to be around say, 40.
P1: They may not always be professors, you know.
Cipher: Okay sir, lets say, 40 panelists. >
P2: Okay, leave it.. gimme the paper you were writing on. Examines it.
P1: So, Cipher, what do you mean by this recession?
Cipher: Sir its about err.. umm... there is this recklessness, and then there is overoptimism about something, and errr.. there is overexpectation from some thing, and then there is a correction about it.. basically goes on in cycles... so the recession is when there is this correction... 
P1: You know you are trying to be very evasive, and trying to get away with it without getting yourself hurt... I think this is what you might have messed up last year in your IIMA interview. Be decisive, and give short answers!
Cipher:  There goes IIMA again.  
Cipher: Sir, lets consider an example, so like lets take the previous recession... .
P2: But then IT is support infrastructure, what do you think of it this time? Why has manufacturing taken a hit?
Cipher: . Sir manufacturing has taken a hit essentially because demand has fallen. And there was also a spike in oil prices in between.
P2: But oils is back now, better than before.

Cipher: Its more about the demand sir, take the example of the big three: Ford, GM and Chrysler, during the good days before recession they were happily working on larger engine sizes, bigger cars, and more cars. More factories, more capacity to feed an artificially high demand. ...
P1: So Can you tell me what is Gini Co-efficient?
Cipher: Sir its in cart... its a kind of a cost function to decide how deep your decision tree should be... I dont know exactly but its a parameter we feed in to the tool that does it for us. 
P1: Looking into my form, says to himself: You guys from Electrical engineering background, doing statistics fulltime... why... 
Cipher: Sir its not that I just veered into it, I generally like data and have done a couple of nice projects on it (Trying to cash in on my btech project and internship which got a paper).
P1: Ho ho ho! So were you the kind of guy who does internships at IIM's during your study at IIT's?? 
Cipher: No sir, this was scientific data on which I had worked, I loved technology, I still love it. No IIM Interns for me.
P2: love tech? Why do you want to do this management thing then?
Cipher:
P2: So you're going off your technology love?
Cipher: No sir it will stay with me. Its an enabler. Will give an edge. 
P1: Some clarification regarding a certificate i received. 
P1: Okay, so tell me, what do you, as a data-miner, do to save the current scenario? What's your client-kind typically?
Cipher: Sir its mostly credit card issuers.
P1: How can you help them in these times.
Cipher: Clumsy again. Sir, err, umm, we can err, umm, help them by reassessing their risk... better analytics to understand how to redistribute credit lines to customers...
P1: THat's done already, and the system is in a mess... they are bleeding right now, how will you help them?? 
Cipher: Sir, Collections! So when the bank writes off or charges off, they still try and recover some bit of it, and some analytics goes in at that level.
P1: Ya, so collections, what happens in that. Quick quick! make it fast! we need to wind up now.
Cipher:
P1: Looks at timer. Okay, leave it. Pick up a toffee.
Cipher: Sir can I complete please?
P1: Naah naah, pick up a toffee, jaao.
Cipher obviously slightly irritated, fiddles around in that small cup to find an eclairs, says thankyou, and walks. out.

-------

Transcript of what I could remember of the encounter at the IIMA interview.