| Sputnikmusic
 

stat_banner

Hello fellow aggregators, and welcome to a celebration of teamwork with data. Rob Mitchum is one such aggregator, and for the past 5 years he has been putting together a spreadsheet per year of all the top 50 AOTY lists released by reputable publications. For this post I did something I don’t normally do, and that’s ask permission to use data, specifically that spreadsheet.

More information can be found on his page, but, essentially, Rob finds all the lists he deems worthy and handscores their ranks into a google sheet. From there, he calculates the average rank for each album (the AVG column), the average rank divided by the number of appearances (WT-AVG column), and a consensus score (CONS SCORE column) which is the average rank of the lists if every non-appearance was given a rank of 75 (an album is thus penalized for every publication ranking it doesn’t appear on). The albums are sorted by that final consensus score and as of January 11, 2018 (which is when I downloaded the data) Kendrick’s Damn. currently stands at number 1.

This very site right here released its own set of end of year album lists so I thought it would be cool to compare our own lists to the individual publications and consensus score rankings. To accomplish this I grabbed the staff top 50 list, the user voted top 50 list, and the rating-based user-usage adjusted top 50 from my year-end rankings post. It wasn’t entirely seamless, I had to change some album titles and band names to fit with discrepancies between the sputnikmusic database (as well as any artifacts from my own scraping of the database) and the AOTY spreadsheet. Another quirk, since there is effectively a loudness war over which publication can release their AOTY list first, I also had to scrape one album from late 2016 (Run the Jewels 3) since it appeared on a few lists for 2017.

(For posterity’s sake, here is the hyperlinked column headers of the AOTY spreadsheet corresponding to each publication’s top 50 album list website link (and the sputnik ones I added): ALTERN, CMPLX, CONSENSUS, CoS, CRACK, DRWND, FACT, GvsB, LINE, MOJO, NME, NOISEY, NPR, PASTE, PFORK, POPMAT, Q, QUIETUS, RSTONE, RTRADE, SKINNY, SPIN, SPUTRATINGS, SPUTUSERS, STAFF, STRGUM, TMT, UNCUT, UPROXX, VNLFCT, WIRE)

So right off the bat, given the consensus scoring, how would the consensus rankings have differed if the sputnik staff list, the user list, and the sputnik user-usage adjusted rankings were included?

new_rank orig_rank change ALBUM
1 1 0 kendrick lamar – damn.
2 3 +1 lorde – melodrama
3 2 -1 sza – ctrl
4 4 0 sampha – process
5 5 0 st. vincent – masseduction
6 8 +2 vince staples – big fish theory
7 6 -1 lcd soundsystem – american dream
8 13 +5 mount eerie – a crow looked at me
9 14 +5 the national – sleep well beast
10 12 +2 the war on drugs – a deeper understanding
11 7 -4 kelela – take me apart
12 10 -2 perfume genius – no shape
13 9 -4 jlin – black origami
14 17 +3 king krule – the ooz
15 20 +5 tyler, the creator – flower boy
16 15 -1 fever ray – plunge
17 11 -6 thundercat – drunk
18 21 +3 slowdive – slowdive
19 16 -3 jay-z – 4:44
20 18 -2 big thief – capacity
21 19 -2 father john misty – pure comedy
22 34 +12 julien baker – turn out the lights
23 22 -1 kaitlyn aurelia smith – the kid
24 27 +3 arca – arca
24 23 -1 migos – culture
26 32 +6 paramore – after laughter
27 42 +15 fleet foxes – crack-up
28 24 -4 julie byrne – not even happiness
29 25 -4 bjork – utopia
30 26 -4 courtney barnett & kurt vile – lotta sea lice
31 28 -3 moses sumney – aromanticism
32 29 -3 jane weaver – modern kosmology
33 30 -3 hurray for the riff raff – the navigator
33 30 -3 j hus – common sense
35 33 -2 richard dawson – peasant
36 35 -1 stormzy – gang signs & prayer
37 36 -1 kelly lee owens – kelly lee owens
38 43 +5 queens of the stone age – villains
39 37 -2 sheer mag – need to fell your love
40 38 -2 jason isbell & the 400 unit – the nashville sound
41 39 -2 japanese breakfast – soft sounds from another planet
42 40 -2 lana del rey – lust for life
43 52 +9 run the jewels – run the jewels 3
44 41 -3 priests – nothing feels natural
45 44 -1 aldous harding – party
45 79 +34 brockhampton – saturation ii
47 50 +3 alvvays – antisocialites
47 45 -2 future – hndrxx
49 46 -3 (sandy) alex g – rocket
50 47 -3 drake – more life

Adding the three sputnik based rankings would have shot up Fleet Foxes and Julien Baker by more than 10 ranks, put The National and Mount Eerie into the top 10, and even brought Run the Jewels and Brockhampton into the top 50 (assuming RTJ even counts).

Another question I had regarding this data was how similar were the rankings for each publication to the others. The following is a plot of the matrix of similarities between each of the publications’ lists as measured by cosine similarity (the similarity metric I used for the musical neighbors post).

cos_sim

In this plot, the stronger the blue, the more similar the publications’ lists are. Note that the top left blocks are greyed-out because they are redundant with with comparisons in the bottom half of the plot. The plot is pretty, but also busy, so what are the top 20 most similar lists in a much more palatable format?

List1 List2 Cosine_sim
PFORK CONSENSUS 0.499
SKINNY CONSENSUS 0.426
SPIN PFORK 0.394
SPUTRATINGS SPUTUSERS 0.393
Q NME 0.371
SKINNY POPMAT 0.352
POPMAT CONSENSUS 0.351
NME CONSENSUS 0.338
SPIN CONSENSUS 0.316
UPROXX CONSENSUS 0.315
STRGUM CONSENSUS 0.304
POPMAT PFORK 0.302
STRGUM CoS 0.298
UPROXX PFORK 0.287
WIRE FACT 0.285
CRACK CONSENSUS 0.279
SPIN POPMAT 0.271
FACT CRACK 0.269
UPROXX NME 0.264
POPMAT LINE 0.261

The consensus ranking (CONSENSUS) comes up quite a bit in this top 20 which makes sense given that it is aggregated from all the other lists found on the AOTY spreadsheet. The Pitchfork (PFORK) list is the most similar publication to the consensus score, it and Spin Magazine (SPIN) are the most similar publications, unsurprisingly the sputnik user list (SPUTUSERS) is most similar to the sputnik user-usage adjusted rankings (SPUTRATINGS), and (though not represented above) the sputnik staff list (STAFF) is most similar to SPUTUSERS and SPUTRATINGS lists, respectively, followed by The Skinny (SKINNY) and the Consequence of Sound (CoS) lists.


Code for this post, including data, can be found here. I may do another post on this data, and/or the data of previous years. Comment if you have any ideas.





macman76
01.16.18
this one seems simple, but it was quite a bit of work

Willie
01.16.18
Crazy shit, this thing you do... I like it though.

Sniff
01.16.18
i don't know if this is interesting or not. Maybe that spin is more like p4k than consensus

macman76
01.16.18
Thanks willie

SandwichBubble
01.16.18
Wire had Circuit Des Yeux in their top 10? Cool

bgillesp
01.16.18
Cool! Seems like Sput users most align with PopMatters? Outside of our staff of course.

bgillesp
01.16.18
What if you added RYM data? Would they be more similar to us or another site?

macman76
01.16.18
I would scrape rym if they let me. If they detect scripting they ban your IP address.

Storm In A Teacup
01.16.18
Paramore got the sput bump ! Yay ! XD

bgillesp
01.16.18
That's lame of them. Just do it on a public library computer then... Just kidding.

macman76
01.16.18
As much work as getting data from hundreds of albums seems, it’s made possible by scripting out the steps, I hit play and in a small amount of time I get all the data. With rym’s rules I would have to point and click my way to all those hundreds of albums. We need to petition the White House for free and open access to rym lol

bgillesp
01.16.18
I'll sign

You need to be logged in to post a comment
Login | Register

FAQ // STAFF & CONTRIBUTORS // SITE FORUM // CONTACT US

Bands: A B C D E F G H I J K L M N O P Q R S T U V W X Y Z


Site Copyright 2005-2019 Sputnikmusic.com
All Album Reviews Displayed With Permission of Authors | Terms of Use | Privacy Policy