music, overanalyzed
honestly this is just me using my data science degree on my last.fm history.. . i'll come back & do this again in december 2026
okay so
june was crazy so ignore.
august is highkey peak..
malcolm todd was 27.8% of everything. that is so stupid...
most new artists did not make it past a week.
so.. how much was i actually listening?
i wanted to see if there was any actual pattern here.
did i actually have range?
august is doing a lot here. june is... not. (i had finals i guess it checks out)
how bad was the replaying?
june is embarrassing bro.
did i ever come back to these artists?
not as much as i thought i did.
where did my listening suddenly change?
this is where the whole thing changed shape for a bit mhm mhm.
when was i always listening?
every play shoved onto one 24 hour clock.
pick someone
random day
okay what happened here
86 scrobbles, 1 artists, 13 tracks, 84.9% repeats
top artist was malcolm todd (quick act shocked)
spin it
drag it around & land on a month.
the nerd bit, if you care
i pulled my last.fm history into pandas & then converted everything to riyadh time first. the weekly line is a 7 day rolling average & the artist spread uses normalized shannon entropy. for the new artist thing i only count people who basically had enough time to reach 7, 30, 60 or 90 days. the taste shift chart here compares each week's artist mix with the week before it using jensen shannon distance. & the weird day is... whichever day ended up furthest from my usual mix of plays, artists, tracks, repeats & variety.