Analyzing Last.fm Listening History | Hacker News Reader

Analyzing Last.fm Listening History

geoffboeing.com

140 pointscobralibre26 comments

This is really cool! I love playing with my own Last.fm data. Some suggestions:

- Manually (or using your Facebook timeline?) place out big events in your life and see if you can see what kind of music you listened to before/after them that you don't normally. I listen to depressing shit post break-ups.

- Use your Twitter or any other feed to estimate your "mood" over time and see if you can find any music you listened to while you're angry

- Use The Echo Nest to get metrics for how happy/sad/energetic the music you listen to is.

- Come up with your own definition of albums played! I hate Last.fm's because, as you mentioned, it heavily biases albums with multiple tracks. In my personal analysis I always start off with a RLE for tracks played on the same album, and my metric for "albums I listen to the most" is "which albums do I listen to at least 3 continuous songs from the most".

I spent a few months this year trying to combine my Last.fm listening history with my geolocation history (I've mostly just stayed in the same place for the last 8 years, 3-4 years) and Forecast.io's historical API to be able to answer the question "am I really only happy when it rains?" -- Data warehousing is a hard problem, though.

263,428 plays over 53,632 songs = 5 plays of each song on average. (updated after comment below)

I haven't "scrobbled" in many years, but checked now that I in 2007 (when I was 16) had 12,377 plays with only 153 unique songs. An average of playing each song 78 times. However, most of the songs I only listened to a few times, so there is basically just ~20+ songs I listened to 250-500 times each that year.

So I guess I was a typical teenager that got obsessed with some songs/artists and listened to them non stop, heh.

This post says there are 53,632 unique songs. 15,503 refers to the number of unique artists.

Shameless plug of a weekend project from a while back (I'm pleasently surprised it still works): http://deja-entendu.zomg.zone

Auto-creates a Spotify playlist from your scrobbles from 6 months, or a year ago or whenever. For instance, if you want to walk a mile in the article author's shoes (ears?): http://deja-entendu.zomg.zone/gboeing

It's kind of stupid, but when I tried out Apple Music, the lack of Last.fm integration really bugged me and I soon switched back to Spotify. Now I understand why :-D

I just installed the scrobbler and it's working fine with Apple Music. FWIW.

been using last.fm with Apple Music since the beginning.

How? Nothing I tried worked with streamed music, only stuff I had locally (especially on iOS)

Are there any scrobblers that work with the iOS version of Apple Music? I tried using QuietScrob, but that requires every track to be in your "library" to work.

From my experience the scrobbler only work with music I own in Apple Music, not any music I'm listening to from Beats Radio or any of the recommended playlists.

On the Mac http://micropixels.pl/neptunes/ works just fine also for streamed music from apple music not just the tracks you download into your local library. Not aware of anything that does it on iOS though unfortunately :(

On macOS, http://bowtieapp.com is a good free alternative.

Original author here. Thanks everyone for the comments! If you've got any other feedback or suggestions I am more than happy to hear it!

This is pretty cool. I still use Last.FM (and also LibreFM) and scrobble from home, work and my phone.

I'll have to give these python notebooks a try :)

Also, your code mostly works on python3. I had to adjust a couple of print statements here and there.

Very cool. I've always wanted to do something like this as well. I'd be interested in analyzing my listening habits throughout the year to see if there is any correlation between months and artists. I feel like I'm pretty "seasonal" with what I listen to.

I tried doing this with my own Last.fm data, if this slow badly coded Heroku app is still live you can see mine: http://lastfm-seasonally.herokuapp.com/user/pvam

Basically, I listen to what I like to listen to no matter the season. If you have last.fm change the last slash in that URL with your username and after a couple minutes (it's single-threaded pulling API requests) it'll show your analysis.

Pretty cool, still sad at the really bad Last.fm redesign but glad useful things come out of the data.

It's pretty awesome what you can do with that dataset. I made a visualization a few years back for a class that looks at the genres I listened to over time:

http://lab.askmike.org/d3/songstream/

That's pretty awesome, very easy to read. I've been meaning to get into d3 soon, what is this particular style of graph called?

Streamgraph

https://bl.ocks.org/mbostock/4060954

It's a Stack Layout turned 90 degrees.

https://github.com/d3/d3/wiki/Stack-Layout

I made something similar with my own data recently. With ~330k plays and 11 years of data, I really enjoyed seeing the long-term changes and trends in my music habits.

I could see where my tastes would change based on where I was in my life (high school, college, entering a relationship, etc). These transitions marked huge changes in my tastes.

It was also cool to see other trends. For example, I found that I have a "concert bounce" where I listen to an artist much more after seeing them live.

This just me realize I haven't been using last.fm for a couple of years, after obsessing on having it in everything I used to listen to music. I guess my move to Pandora didn't help.

But what do I know, there's a browser extension that does that! http://build.last.fm/item/1000591

Back to last.fm I go.

If you look at my listening history, Untitled is by far my favorite track. That's because every interlude on one album is untitled so I end up listening to it like 5 times every time I listen to the album. Never mind the track number, though, it doesn't look at it.

Cool I have a great collection in lastfm also. So I should check it out...

Reply on news.ycombinator.com