Econometrics vs Machine Learning

Machine learning (in particular deep learning) is currently a hot topic in CS and the social sciences because it can uncover reularities in large, high-dimensional datasets (and it really seems to work).

What is the relationship between machine learning and econometrics? I highly recommend two recent talks.

1. Hal Varian: Machine Learning and Econometrics (Youtube)
2. Sendhil Mullainathan: What Big Data Means For Social Science (edge.org)

What to demand from a Scientific Computing Language (even if you don't care about computing or languages)

"There is something in parantheses that is turning peoples' brains off." The great Peter Norvig (from Google and our favourite AI text-book) in defend of the Python programming language.

http://www.msri.org/summer_schools/556/schedules/4197

Should you outsource analytics?

Can the market provide better analytics than your org? A very interesting piece in the MITSloan Management Review. Read it here.

Introductions to Big Data Analytics

There are now a couple of introductory textbooks for Big Data analytics. In a recent post the MIT Sloan Management Review discusses three of them: 1) Big Data: A Revolution That Will Transform How We Live, Work, and Think (Houghton Mifflin Harcourt, 2013), 2) Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die (John Wiley & Sons, 2013) and 3) Keeping Up with the Quants: Your Guide to Understanding and Using Analytics (Harvard Business School Publishing, 2013). Click here to read.

I would like to add the great "Data Science for Business: What you need to know about data mining and data-analytic thinking" (here). Hal Varian, chief economist at Google has distilled some thoughts about Big Data and econometrics into this piece.

Students interested in such topics should consider our Advanced IS course and the network science course! See you there!

What scientific idea is ready for retirement?

The annual edge.org question for 2014: WHAT SCIENTIFIC IDEA IS READY FOR RETIREMENT? What is cool about the edge is that you can either get the book or just read the answers online (here). Like every year it is quite difficult to spot a single answer from a German scientist. Anyway, there are surely some thought-provoking answers, so check them out.

Friends change, but their number may not

I just stumbeld across a nice paper that combines cell-phone records and surveys to understand how friendships change over time and how many close friends we have. The results "[...] are likely to reflect limitations in the ability of humans to maintain many emotionally close relationships, both because of limited time and because the emotional “capital” that individuals can allocate between family members and friends is finite."

So, Facebook is one thing, real life is another. Ok, this may not be spectacular news for many, but scientific research into the differences between online and offline social network formation is rare, and this is why the paper appears to be interesting.

Persistence of social signatures in human communication, in the Proceedings of the National Academy of the Sciences (PNAS)

A short introductory discussion on Sciencemag.

Why IS researchers should study networks

Many disciplines study social and information networks. I have always felt that IS researchers are in a somewhat special position to do this. They should know how to collect data, they mostly know graph theory, the can code, they should know stats and (maybe most importantly) they build networks, or software that reveals networks. There is still little network science research going on in IS/BISE, or at least, it could be much more! Maybe this very nice paper by Sundararajan, Provost, Oestreicher-Singer and Aral changes that.

A must read: Information in Digital, Economic and Social Networks , forthcoming in ISR.