Monday, September 22, 2014

"Big Data" is a Religion

As far as I can see, "Big Data" seems more like a religion than anything else.

I like "Good Data" (which is far from the same thing) and "Good Models" (which can, e.g., include parameters fitted from data). I feel like I have a chance to actually understand things that way.

Data analysis definitely has its place and I certainly also have my moments when I do data analysis (see some of my papers), but data is far from everything. In short, I have ambiguous feelings about "Big Data".

Anyway, data analytics is an approach, and I use it sometimes, but there are also many other approaches that one should pursue to try to understand phenomena and solve problems.

By the way, I first posted the picture to which I link above in this blog entry, I ranted a bit about it in this blog entry, and I talked about a Small Data project at ECCS in 2013. (That work I described in that talk has subsequently been published in PNAS.)

2 comments:

Krishna Bhogaonker said...

Haha, I agree with you. As a statistics person, "Big Data" just means more of our money goes to the Computer Science departments from Stats. Yes, I did go there :). But more than that, I think that this is again a case of people taking a simple tool and applying it in inappropriate ways--just like linear regression. I just wish more people were working on the underlying social/economic/behavioral theory rather than purely data analytics. P.S. I am looking forward to your talk at UCLA in October :).

Mason said...

See you next month! (There will be some data, but this is definitely going to be a "small data" talk.)