macman76: Resorting the Top 200 2: User Usage Weighted Mean

macman76 05.28.16	User Usage Weighted Mean is on average .08 rating points less than the standard mean rating Also, I left space so people could try and find albums that could rank higher with this new rating system... so guess away.
MuhNamesTyler 05.28.16	Holy shiiit
macman76 05.28.16	check grotesque
Alastor 05.28.16	Hemispheres won
macman76 05.28.16	not listed, but Camel-Moonmadness goes from rank 196 to rank 66 with the new rating
Sinternet 05.28.16	goes to prove userbase is wank
macman76 05.28.16	this rating system is different from the old one, guess if an album would make the top 101 and I'll post it, I did aspid and grotesque as an example
ArsMoriendi 05.28.16	Hot Rats?
Alastor 05.28.16	you need to find an algorithm where blood fire death is 1.
macman76 05.28.16	@al hater-weighted mean
DoofusWainwright 05.28.16	Which user uses the site the most?
macman76 05.29.16	by my weighting, number one is BallsToTheWall, mainly because he's a huge outlier in lists. however, I adjusted each user usage value so that it was a percentage of all for each metric. Without that adjustment, the number one is KILL.
ArsMoriendi 05.29.16	Mr. Bungle - California
parksungjoon 05.29.16	"Without that adjustment, the number one is KILL." hahaha thrash lives m/
macman76 05.29.16	mr bungle was .02 rating points away from making the top 101
ArsMoriendi 05.29.16	I figured :P
onionbubs 05.29.16	cure - pornography joy division - closer
macman76 05.29.16	got you bubs
parksungjoon 05.29.16	demilich - nespithe aspid - extravazation
hal1ax 05.29.16	how does this algorithm compare to potsy's algorithm
macman76 05.29.16	@hal, yes it does its a collaborative filtering algorithm (wiki it if you want more in depth info) that takes inputs of user rating information and can be used to predict what a user would like (specifically how they would rate something) given the behavior of other users. I modified my dataset (all user ratings for all staff reviewed albums, 7821 albums total, scraped circa March 20th) and used code from this page http://stackoverflow.com/questions/1022649/weighted-slope-one-algorithm-porting-from-python-to-r to calculate slope one (I believe it is the weighted slope one) recommended album ratings. I narrowed it down to users with 40 or more ratings (12,516 users) and calculated the (1) correlation of the users ratings to the average rating of the albums they rated, (2) the average number of rating per album rated by user, (3) the number of ratings by the user, (4) the number of ratings per day by the user, and (5) the standard deviation of their ratings. All values were transformed into rank values and a weighted average of the ranks determined you troll rating. For Contrarian Troll, a weight of 20 was given to the correlation, .5 for the average number of ratings per album, 2 for number of ratings, and 1 for ratings per day (weights were determined by a process that can best be described by the shrug emoji and Rachel Ray’s philosophy to measuring ingredients). Presented, also, is the average rating of the album, adjusted mean, percent of users that had positive adjustments (meaning that their average rating was higher than the average rating of the albums they rated), and the median number of albums that users of that album had rated within this dataset of 7821 albums. After that, the top 10 and bottom 10 albums based on adjusted rating are presented in the forms of Beyond Omega, Omega, Alpha, Beta, Gamma, and Feltchsilon. ... but for albums
hal1ax 05.29.16	lol
parksungjoon 05.29.16	"of Beyond Omega, Omega, Alpha, Beta, Gamma" smh plebs don't know greek alphabet, omega is literally the last letter bro that's like getting a Z grade in school or some shit
parksungjoon 05.29.16	for real tho good job this is really cool
hal1ax 05.29.16	don't argue with the algorithm u will b destroyed
parksungjoon 05.29.16	oh shit
macman76 05.29.16	I know how to count ... one, beyond omega, omega
macman76 05.29.16	Bump
parksungjoon 05.29.16	dismember - like an ever flowing stream
macman76 05.29.16	got you park
lobstah 05.30.16	Great list although you lost me as soon as you said "square root".
macman76 05.30.16	@lob You're gonna love this then... logarithm
Winesburgohio 05.30.16	christ
iloveyouall 05.30.16	did you know algorithm and logarithm are anagrams of each other!?!?!!
hal1ax 05.30.16	WOAH
macman76 05.30.16	A Love Ill You Chas
brainmelter 06.01.16	Peste Noire La Sanie des siecles - Panegyrique de la
macman76 06.01.16	got ya
brainmelter 06.01.16	nice its 66
macman76 06.01.16	Actually, no... The request ones aren't in a particular order, but I denote if they would have made the top 101. It can still be 66 in your heart though
parksungjoon 06.01.16	immolation - here in after
brainmelter 06.01.16	in my heart yes.
TVC15 06.30.16	Bump because I miss this guy :[
evilford 02.28.17	3 albums in the top 7 what up now floyd haters