Is it Music?

We can see that the tempo and key signature of a song don’t have anything to do with popularity. However, we looked for a few more musical attributes and their correlation to song popularity.

What is Spotify doing? What does this mean?

We think that this might come from the fact that Spotify uses the natural popularity rating to help suggest songs to its users, and they need to manually alter some songs so they don’t get recommended to listeners. For example, the Ghostbusters theme song has a popularity of 0, even though it was the 9th most popular single of 1984. For reference, the 10th song in 1984, “Karma Chameleon”, has a rating of 78. It could be possible that Spotify shot down the popularity rating of the track because they don’t want people to be unhappy to hear Halloween music year-round, especially when it’s being recommended with other popular songs. Another example is Whitesnake’s hit song “Here I Go Again” which has a popularity of 0 while being the number 7th hit song of 1987. Spotify released a remastered version of the track in 2018 and probably wants that specific version of the song to be recommended to its users instead of the original recording. When analyzing things like music, it’s important to realize how the industry can control certain aspects of how it works, and how it can skew your data and research.

First Category

All of these songs use language about losing someone and wanting them to come back into the narrator’s life. However, it still uses the theme of heartbreak and love that was present in the first two categories.

Second Category

The second category is extremely similar to the second category designed by the first model. Most of the songs highlight the excitement of being in love and sharing positive feelings about another person.

We can also see that “I Look Away” by Chicago appears on both of those lists. This song is very sad, and should probably fall into a different category, but the topic model sorted it with all of the happy songs about being in love. This is because it uses the words “Happy” in the sentence “I’m really happy for you” over and over again when referring to what the narrator said after he was informed he’s been replaced in a relationship. While this note isn’t as important in determining the types of popular songs, It’s worth pointing out that the topic model is prone to errors as It can’t understand context and language very accurately, potentially messing up your results.

Third Category

The third category stands out a bit, as all of these songs seem to lack compatibility. Even though they are all written with themes of love or heartbreak, just like the rest of the songs, they lack a specific trait relating to those feelings that they all share. The thing that could potentially hold these songs in a group is that they all use repetitive words or phrases throughout the whole song. “…Baby One More Time” by Britney Spears uses the word “Baby” 26 times throughout the whole song, and “Happy” by Pharrell Williams uses the word “Happy” 36 times. “What Makes You Beautiful” uses a series of “Nahs” throughout the whole song as well. This could mean that another possible characteristic of popular songs is word or phrase repetition.

Fourth Category

Very similar to the first category of the first model with only two topics, this group of songs all use language about heartbreak or negative feelings towards a person. As mentioned before, the topic model does not always do an effective job of recognizing tone. We can see another example of this with “Wake Me Up Before You Go-Go” being on the list. This song is very much a happy song about wanting to join somebody to go dancing, but the language used suggests something else. The line “Something ain’t right, My best friend told me what you did last night” could be picked up as being heartbreaking or negative.

The 4 category topic model didn’t provide much more insight into the subject, proving that most popular songs are written about either love or heartbreak.