One of my favorite times of the year is December. The lights, the cold weather, the festivities, and Spotify Wrapped! It’s something my friends look forward to at the end of every year as we love to see who are favorite artists are backed by the data of our actual listening habits. While we wait for Spotify Wrapped to come around each year, my friends and I love trying out new and different sites that that give a high level overview of our relationship with music such as Stats for Spotify, Obscurify, or having an AI judge my music taste. However, Spotify Wrapped included, they all hide the magnitude of the data from you (not to mention Spotify Wrapped excludes November and December from its charts). It’s one thing to know your who your favorite artist is, but to see you listen to them twice as much? A lot more exciting!

There is a lot of nuance that is missed about how I listen to music despite the fact that Spotify has all the data.

1. Dataset

Last.fm is a popular service that you can hook up to your streaming services to automatically record what songs and artists you listen to. Scrobbling is a term that means an online service is recording data of a user’s listening history, with Last.FM being one the most popular. In a similar fashion, a scrobble is one record or instance of someone listening to a track. For example, I have 10 scrobbles on a song means I have listened to that song 10 times. I used an online tool where you just give it your username and it exports your Last.fm data to a CSV file (along with other options).

In Tableau, I used groups to clean up some minor data inconsistencies for some of my most listened to artists. For example, songs with featured artists are often listed as “Artist1 & Artist2” rather than attributed to all the individual artists. To simplify this, I only attributed the artist who released the track to make sure each scrobble of a song was only counted once. While this is slightly inaccurate, featured artists are not a common feature in my music and would only account for a fraction of a percentage of my scrobbles.

2. Inspiration

My inspiration for this comes from two main sources.

I knew that I wanted to make some new visualizations that gave a different aspect of how I listened to music. Something that was a little more personal and differed from how the other sites I previously mentioned looked at music data.

When searching Spotify visualizations I came across Yaron Yitzhak’s guide on how to visualize Spotify data with Google’s Data Studio. I liked Yitzhak’s visualization because it visualized the data compared to the lists I am used to and that it gave some summary stats at the very beginning. However, for my visualizations, I wanted to experiment with graphs other than bar graphs.

3. Making the Charts

I went about making the graphs by exploring the data in Tableau. Since this was my first experience with Tableau and I enjoy learning by doing, I went to Tableau and started throwing data on every axis to see what I could create. I started with simple bar charts listing my favorite songs and artists, but quickly explored how I could add time into the mix. I ended up making about 16 graphs, with some exploring the best way to present the same data.

During this whole process, I tried to think about how I listen to music and what is not represented in the data and graphs I already have available to me. I tend to obsess over songs and artists for periods of times, listening to albums or songs on repeat for days or even weeks. When I look at my all time favorite artists, I am sometimes shocked about how newer artists are able to make their way up the list so quickly. So I decided to visualize that by tracking my favorite artists scrobbles over time. In a similar fashion, some artists I love have been making music for a decade and have discographies of over 50+ songs while other artists have been releasing music for under two years and do not even have a debut album. I have always felt putting artists on the same all time favorite lists does not fully represent the whole picture which inspired so I wanted to compare the amount of unique tracks I have listened to of an artists compared to how many scrobbles I have accumulated for that artist.

4. Results

From making these charts, I was surprised to see some trends.

1. Olivia Rodrigo shot up fast compared to the other four artists who has been a constant since I started my Last.Fm account.
2. While it is no surprise to see that I listened to music a lot more in quarantine than in the last year, it’s an aspect that is not well reflected in other music stats platforms.
   3. I knew Steven Universe was my number one artist by quite a bit, but I did not know it was my recent binge listening that pushed it so far above.

5. Reflection

As my first time using Tableau, this experience was fun yet a little difficult. I found myself still trying to figure out little details about Tableau like how to filter data exactly how I want it or formatting. While I do not believe my visualizations are perfect and the formatting could be better, I am happy with the results as my first time user.

Limitations

1. Inconsistent and false scrobbles. In the past, I used a web scrobbler browser extension to track my scrobbles; however, found the extension unreliable in producing data consistent with Spotify and it was tracking regular YouTube videos so turned it off. Despite it gone, I never deleted the incorrect data from Last.fm. Though, these hiccups should have no big effect on my visualizations since I am focusing on the most listened artists and songs.
2. Spotify data did not come soon enough. Ideally, I wanted to use my extended streaming history which would have had all my listening information on Spotify since the creation of my account, more dimensions such as time spent listening, skips, etc. and most likely would have been more accurate. I requested it, but due to the process taking up to 30 days, I was not able to obtain my data from Spotify in time for this assignment.
3. No reliable unique identification of songs, artists, and albums. While the tool I used tried to give each song, artist, and album a unique identifier, I found the tool left too many blank cells to reliably use for data analysis.

Future Directions

1. Include genres. In the future, I would love to include genres. I would love to see how my preference for genres changes (if any) over time.
2. Time spent listening. An aspect included in Yitzhak’s visualization is how long he spent listening to music. While not included in the Last.fm dataset, how much you listen to of each song is included in the data you can get from Spotify.
3. Better aesthetics. As I get better with Tableau, I would love to make my visualizations more representative of me and my taste in music. I tried to keep it simple since this is a new tool, but I would love to have a little bit more creative and artistic visualization.

References

Bowenbank, S. (2021, November 10). Spotify Wrapped: Here’s How to See Your Top Music For 2021. Billboard. https://www.billboard.com/business/streaming/spotify-wrapped-insights-how-to-find-2021-9657709/Definition of scrobble | Dictionary.com. (n.d.). Www.Dictionary.Com. Retrieved March 3, 2022, from https://www.dictionary.com/browse/scrobbleThe number of times my kids said “mom” in the span of a week. #IronQuest May 2020 #quantifiedself entry. (n.d.). Tableau Software. Retrieved March 2, 2022, from https://public.tableau.com/views/AWeekofMoms/Mom?%3Adisplay_static_image=y&%3AbootstrapWhenNotified=true&%3Aembed=true&%3Alanguage=en-US&:embed=y&:showVizHome=n&:apiID=host0#navType=0&navSrc=Parse Yitzhak, Y. (2020, June 19). A simple guide to visualizing your Spotify listening data… badass-ly. TNW | Tech. https://thenextweb.com/news/a-simple-guide-to-visualising-your-spotify-listening-data-badass-ly