Modern Metrix

What Can Your Electric Meter Tell About You? Your Favorite TV Shows!

For those of you who can speak (or read) German, this interesting paper discusses a study on how analyzing data from advanced electric meters can reveal which TV shows are particularly popular in your house.

 

Mobile web as a threat to online survey research?

I noticed lately that more of my friends (and myself) stay away from their computers and laptops and prefer to access the "essential" web functions like news, e-mail and social networks via apps on their mobile phones, tablets and devices like Roku or Xbox. In my opinion, this could pose a serious threat in the future to online survey research industry, similar to what cell phones did to the telephone surveys nowdays (reach, non-representativeness, lower response and cooperation rates, etc.)

What's up with "Big Data"

In her recent post "Bid Data? Big Deal" LoveStats rightfully pokes fun at the term "big data" that became a new buzzword of social media folks. As she correctly points out, there is nothing special or new about big data sets; they have been around since Census data became available in digital format. Unfortunately, more than often this fascination tends to be with the size and "messiness" of such data, and rarely with their content and potential value. True, so many businesses and organizations now have accumulated thousands and millions of data points on their customers, staff, etc. It is also true, most of them have no clue what to do with this information. The common notion is that these untouched "gold mines" are just waiting to be ..yes, mined (as in data mining). However, most of the time when the term "big data" is mentioned, the main emphasis is usually on data organization and management (how to restructure that data and keep it up to date) and rarely on extracting insights and analytics. It's all about storing (cloud! cloud!!), restructuring, and visualizing (dashboards!!!!).

 And that makes me a sad, sad panda...

SAS IML cheat sheets

As a huge fan of matrix language and SAS/IML module, I really appreciate these nice "cheat sheets" created by Rick Wicklin, developer of IML procedure at SAS Institute.

Google's Ngrams project

Just stumbled upon Google books "Ngram viewer", a time serialized dataset of words and phrases from over 5 million books in seven languages that were printed in the past few centuries. What a treausure trove for crypthoanalysts!

Tweets and biological rhythms

NYT discusses a paper by Scott Golder and Michael Macy from Cornell University on correlation between the emotional tone of people's messages on Twitter and known biological rhytms (aka cyrcadian rhytms). Using text mining software and impressive sample (several millions of tweets), researchers classified random tweets based on their content (usage of words "awesome", "great' and emoticons) as either "positive" or "negative", and then looked at the patterns throughout the day. They found that for majority of people, positive tweets tend to appear in the morning (between 6 am and 9 am), slowly decline towards afternoon, and then rise again after dinner. In the beginning of the workweek the overall mood tends to be lower, with the curve peaking on weekends. The patterns seem to be the same across different cultures.  

 

Image source and more info are here