Monthly Archives: July 2019

Blog Post 2 Redux

I chose to redo this blog post since I worked on it a week where I had NO time. I ket the visualizations the same for the most part and spent more time cleaning the data than anything else. I used the entire music library that is on my laptop.  make that distinction because my iTunes are split pretty equally between an old I-Mac and my laptop. I got this laptop in 2017 when my old one got wet and fried ( R.I.P.!)  and started it with a backup of that computer. I got my first laptop in 2011.  My new laptop did not have the same amount of space as the old one so I ended up with my two iTunes libraries. The dataset I used included 2349 songs.

Originally I wanted to answer the question “how diverse is the music I listen to?” I thought of diversity geographically and wanted to map out the artists in my library by birthplace. I was interested in where my music was coming from and how music and culture can cross international borders. That topic was too big for the time contraints, but my tiny initial inquiry proved to be fruitful in answering that question. A dance/electronic song in Japanese that I enjoy was made by a young woman in Brooklyn, and a few Afrobeats artists I like are London born and bred.

I kept this in mind while cleaning my data. I spent the most time working on the genres. I had many songs with the same genre that was spelled incorrectly or all-caps while others were sentence case. I spent a lot of time going through and making sure the genres were uniform. I made a decision to only make changes to songs that already had a genre. If the song did not have a genre I left it blank, because as it showed in my first version of this project the null values told the story. I think the nulls show the change in technology over time and changes in the ways that we acquire music. In the mid to late 2000s most of my music was imported manually whereas now much of my music new music is added at the click of a button. In addition, it appears with the title, artist, genre, and much more.


A similar feeling happened when I was working on the genres. First was my conundrum on how to divide rap, hip-hop, and R&B. These genres are often used together to describe something.  I also listen to a lot of alternative R&B/Soul music that might be a mixture of all three of those things. I decided to separate hip hop and R&B because I felt that having a hip hop and R&B category in addition to  Hip-Hop / Rap and R&B / Soul would be redundant. I then had to decide if I was placing a song in a certain genre because of the artist or because of the song itself. Different artists dabble in different genres and try new things, which made it a bit difficult to choose  who goes where. I decided to go by the song which then allowed the genre to give some context back to the artist. Some artists came up in more than one genre and I left them in those different genres. For example Goldlink is a rapper /soul artist, so one of his songs may show up in the R&B/ Soul genre while another song shows up in the Hip-Hop/Rap category.

For the bubble chat below I decided to show the total number of plays per genre. I deliberately left the null values in the chart and used color and size to make it stand out.  


The second thing I spent a lot of time cleaning was the artist field. I wanted to make sure that the artist name in each field was the main person on the track, not including featured artists. As I was doing this I could feel myself stripping away the depth some of the songs. Collaborations between artists of different genres became just the artist who came first on the list.


The third thing I wanted to show was the change in my taste over time. That was one of the charts I wanted to make, but struggled with in my first try. To do this I used a dispersion plot to show when certain genres are added to my iTunes library and when they stop being added. The genres are order by the total amount of plays. My last try Easy Listening came up as a shocking top genre I listened to. Here it’s a bit lower on the scale. That being said Easy Listening is comprised of one Andrea Bocelli album. That album is the most played album in my entire library! I think this shows my habit of listening to him as a way to calm down before bed at night or as a relaxer during my morning commute. I left the null values in the chart because again I felt they told the story.  The nulls are at the top of the list. In the tooltips, I included the number of plays of that genre from the time it was added to my Itunes library to the day I downloaded the sheet (June 9, 2019). It was interesting to look at the dramatic drop in the null values. To better show it I made a line chart of just the null values.


In 2013, songs without a genre had over 6,000 plays and by 2018 songs without a genre had only three plays. I think this speaks to what I mentioned earlier with the changes in how we consume music. I though the use of streaming services to add music to my library as opposed to adding things in manually was a factor in the decrease of nulls. I tried checking my Apple account to find out  when I first signed up for Apple music.Apple doesn’t allow you to see when you started the subscription, but the first Apple music receipt I found in my email was sent in 2017. By then most of my new music was coming in with genres, so I don’t think Apple music truly correlates with what I’m seeing in the line chart.