Data analysts love music, don't they?

any good examples?

*examples of good data analysis on music

https://www.youtube.com/watch?v=kPRA0W1kECg

lol Ozzy Osbourne in that pic what is this 2003

also they're just picking on Melanie Fiona damn

i once saw a chapter in some math book where they programmed an algorithm to complete an unfinished Beethoven symphony but have never been able to find it since

Like any data analysis / research, it's only as good as the researchers can make it. Reading levels is bullshit, but there are plenty of interesting ones out there. Always vaguely dubious of the one where people are brazenly trying to predict what will be popular next based on what was popular last, though. That goes for everything, though, not just music.

hi! data analyst here.

first, no, we don't have anything better to do. i mean, we do, but feeding taylor swift songs into a computer makes a nice break from analyzing effects on efficiency of ehr implementation.

second, this is crap data analysis, and it only makes it into newspapers because apparently people don't know how to look at charts. i mean, look at that chart of "genre intelligence comparison over the last 10 years". what does that tell you? all i see is a bunch of lines oscillating pseudorandomly between the values of 2 and 4. assume a margin of error of one grade level, which is pretty reasonable all things considered, and the obvious conclusion is that there has been no change at all in word complexity over the past ten years.

see also: that horrible chart that's been floating around linking bands people like to sat scores (which i suppose is more of an american thing). pseudo-scientific claptrap.

that's not even mentioning other questionable assumptions in the article, such as that louisville, a city where i lived for a decade and still visit regularly, is somehow a haven for country music. also, of course, the entire notion that artistic validity is determined by grade level is wholly undermined by "green eggs and ham".

Does the line shoot upwards during the time period that "O Superman" was in the top ten?

also, if he's that concerned about city names skewing his data, why didn't he just set his calculations to exclude proper nouns? i mean, seriously, who taught this guy how to analyze data, martin rimm?

Seems to me that a very general problem in data-analysis is that if you analyze data you don't know that much about, any results seem as reasonable as any other results.

would love to do a geir hongro inspired "# of chord changes per song" time series

i think there was a thing a short while ago like that.

yep http://matthiasmauch.de/_pdf/MauchEtAl_EvolutionPopUSA1960-2010.pdf

I am sympathetic to the idea of playing with data and seeing what falls out but I can confidently say that almost all studies purporting to rely on deep linguistic analysis of lyrics are going to be nonsense.

this is kinda cool http://j.mp/1LfobQG

also kinda 'thx for the email dad'

What drives me crazy about things like the Mauch study is that there is plenty of equally-sophisticated analytical work on popular music that is done by people who actually have expertise in the subject, that deals with much better questions than "whether pop musical evolution has been gradual or punctuated", none of which ever gets written up in The Guardian.