x
Big Data

Telecom Analytics Grows Up

The big data analytics debate has moved on from a year ago, with some experts suggesting it's no longer a technology challenge.

Ariella 11/17/2014 | 5:50:46 PM
Re: there's a whole lot of useless data... @mhf1ve I love the way that is described as "The data point no one asked for!"
mhhf1ve 11/17/2014 | 5:29:37 PM
Re: there's a whole lot of useless data... Yes, there are an endless stream of examples for correlations that have no meaning:

http://www.tylervigen.com/

Big data doesn't solve this problem, and if anything, it makes it worse for finding more and more spurious correlations that don't really have any causal relationships at all. Retailers might get excited to see some correlations between various things they can measure, but there may not be any reason behind them to keep the trend going. On the other hand, there might be..

http://time.com/3583945/alibaba-singles-day-bra-sizes/

Alibaba has linked women's bra size to their online shopping habits—and it found that the bigger the cup size, the bigger the spending.

Ariella 11/17/2014 | 2:48:47 PM
Re: there's a whole lot of useless data... "Mathematicians can correlate a lot of variables, but the *causation* aspects and designing the controlled experiments to figure out if those correlations have any meaning is where the "data science" attains value."

@mhhf1ve that's exactly the crux of the issue. I recently saw a graph that pointed out correlations that really have no bearing on ascertaining causes.   There are some examples in this: romquarkstoquasars.com/correlation-vs-causation/
mhhf1ve 11/14/2014 | 4:24:33 PM
there's a whole lot of useless data... Big data has been around for a while -- depending on how you define "big" -- and just because we can collect an insanely large amount of data doesn't necessarily mean all of it is going to actually be valuable or actionable. Mathematicians can correlate a lot of variables, but the *causation* aspects and designing the controlled experiments to figure out if those correlations have any meaning is where the "data science" attains value.

That's where the subject matter expertise comes in... and where the bottleneck of human intelligence lies. Computational power is not the limiting resource in big data productivity.
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE