Information

So last night I finished up the details for my volunteership at the hospital. On one of the forms, they asked for my birth date, full name, social security number, driver's license number, home address, and phone number. Besides like my pants size, they knew everything about me. I was really nervous, because in my data mining class, we're going over the proper use of data.

This morning in class, we talked about how much information should be gathered. I had a couple thoughts on that:

With regards to terrorists, law enforcement officials have a huge problem with the whole data mining thing because if the computer goes through the data and tells them to postpone this flight and stop the man with the suitcase, they must do it. If they don't do it, they could be held accountable if the man was really a terrorist.
I thought this was interesting, and wondered what makes people trust computer so much. It wasn't only a data mining activity that predicted the 9/11 terrorist activities. It was also Nostradamus, and even folding the twenty-dollar bill in half, and then folding it around... people should have been able to look to their currency and prepare for the terrorist attacks! (see http://www.foldmoney.com/)
So what makes the predictions given by computers so influential? Perhaps it is grounded in the magical mystery associated with computers.

Of course, this doesn't just apply to terrorists. Our data mining teacher had us read an article about how doctors don't want to be subjected to data mining (see http://www.iht.com/articles/2006/05/04/business/Drugs.php). They don't want their prescription patterns analyzed. Data collection could also eventually lead to a computer giving a recommended prognosis, and the doctor choosing to follow or ignore the prognosis. This would place doctors under the same liability that law enforcement officials are placed under. If they fail to follow the recommendation of the computer and the patient dies, are they accountable for making the bad decision?

I wonder what the solution to these problems is. Perhaps the ugly but true answer may be more information disclosure. To avoid getting stopped at airport security, we may have to enroll in a program similar to FlyClear, (see http://www.flyclear.com/what/) who gathers a ton of information. According to the website, you should have the following items handy during enrollment:
* Driver’s license number.
* Home address for the past five years.
* Social security number.
* Alien registration number (only for permanent foreign residents)
* A valid credit card.
And then "In person, we’ll verify two pieces of approved government issued identification, capture images of your irises and fingerprints, and take your photograph."

If it didn't cost $199 bucks a year (that's like half my annual income), maybe I'd do it.

If everyone except terrorists had a FlyClear card, it would be easy to track them. But I don't imagine everyone will be ok with divulging this much information. Leave it to the granolas to spoil the anti-terrorism movements..

Grad schools

I had a good talk today with Dr. Johnson from Stats about grad schools. Incidentally, the talk happened during my civ class.. but it sounds like I didn't miss anything. Nathan says that he just pontificated the whole class.

Anyway, Dr. Johnson brought up a couple of good points. If I stick with a bioinformatics degree, it will be hard to find a faculty teaching position when I graduate. If I graduate from a biology department, I will also probably have to do a post-doc before I can get hired on anywhere.

This was really good for me to hear. I'm going to have to look into more cs-ey programs. I have always only considered applying for bioinformatics programs, and assumed that my research experience and specialized undergraduate degree would help me get my foot in the door there, but I'm not sure what advantages I will have when compared to other CS students. Especially with the economic recession and people going back to grad school, there will probably be more competition from professionals returning to get a PhD. Perhaps the computer science industry won't be so hard-hit. I haven't felt any less demand for CS graduates in industry, and the same recruiters keep coming to recruit new graduates. I guess we'll see, though.

Gibbs

So I took this Bayesian Statistics class last semester. At 8:00 in the morning. It was killer.
Anyway, today I was doing some stuff in R, looked in my workspace, and realized I still had a Gibbs Sampler method, affectionately named GillsGibbs after my professor. So if any of you are looking for a R function that does Gibbs sampling, have at it:

function(mu0,sigma02,shape,scale,data,n)
{
 ybar <- mean(data)
 out <- matrix(0,n,4)
 w2 <- 1/sigma02
 sh <- shape + length(data)/2
 sigma2 <- 1/(scale*(shape-1))
 for (i in 1:n)
 {
  if (i == 1) {w1 <- length(data)/sigma2}
  else {w1 <- length(data)/out[i-1,2]}
  postpre<-w1+w2
  postvar <- 1/postpre
  postsd <- sqrt(postvar)
  postmn <- (w1*ybar+w2*mu0)/postpre
  out[i,1] <- rnorm(1,postmn,postsd)
  ssy <- sum((data-out[i,1])^2)
  sc <- 2*scale/(2+scale*ssy)
  draw <- rgamma(1,shape=sh,scale=sc)
  out[i,2] <- 1/draw
  draw1 <- rgamma(1,shape=shape,scale=scale)
  draw1 <- 1/draw1
  draw2 <- rnorm(1,mu0,sqrt(sigma02))
  out[i,3] <- rnorm (1,out[i,1],sqrt(out[i,2]))
  out[i,4] <- rnorm(1,draw2,sqrt(draw1))
 }
 return(out)
}

It's pretty cool.
However, I do have one quick rant: Why does the <code> tag not preserve indentation?

Gurlz

So, I'm sitting in my 252 class, and this kid behind me said that he was asked on a date to go to the homecoming spectacular. Of course, this kid is a boy.. it's CS 252. Anyway, somehow, this kid was asked by his date if he wanted to pay for the tickets. I had to laugh... girls these days.

Boston

Boston has been a blast. It started on the plane ride over. I met this guy named Dan, who was a chemical salesman from Boston. AmGen, I think. He told me that a critical part of being a salesman is offering suggestions. He noted that when you compare prices and products at different stores, you are most likely to buy something from the salesman who tells you "what you really need is something that does this, and this is the product that does that." Instead of someone who says "yeah, I think that is a good choice." I thought that was interesting.
Here in Boston, I have gotten excited about research, and realized that a lot of people have no idea what they are doing, and just are publishing what they have found. So I think I'm on the right track with my having no idea. It comes naturally for me. I also saw some interesting approaches to things, and was inspired to try some different stuff when I get home. Things that impressed me were the way people handled the changing of pathways throughout developmental cycles, and how they used a combination of techniques to conquer problems.
Today we got tired of the conference (it's been three days of listening to lectures from 9 to 10 with short coffee breaks and an hour lunch with feta cheese) so we decided to see a little bit of Boston. We saw Bunker Hill, the Old North Church while listening to Dad read Longfellow, and even saw the MIT museum which was really interesting and motivating.