Gimme, Gimme, Gimme (An Analysis of ABBA)
Introduction
A lot of people have asked me, "how would you rate all 112 ABBA songs?" I eventually got tired of the question, so I decided to answer everyone all at once. Over the course of a few weeks, I listened to all 112 ABBA songs, as defined by the wikipedia page "List of songs Recorded by ABBA" and rated each of them on a scale of 0-5. Additionally, I recorded the number of Spotify plays of each song at the time I listened to them. I then did some digging into the data I created. So without further ado, let's get started!
Rating Breakdown
As you can see I generally think ABBA is pretty good. The average score was 2.93, which corresponds to just below a solid score of three. I only gave a single song a 0, but I think it deserves it.
Average Score and Total Plays by Year
This isn't incredibly surprising, but it looks like ABBA was at their best in the late seventies and early eighties. Additionally, their return album in 2021 was pretty good score wise, but not nearly as popular as their early songs.
Their best year according to score was 1978. However, that was only a single song that I happened to like, so it's not the best conclusion to draw.
Popularity
I was curious to see how my tastes aligned with the broader public's. To do this I plotted my scores vs spotify plays.
This didn't reveal any relationship to me, so I decided to plot my scores vs the log10 popularity of each song. Basically, we're going to be looking at how many zeros are in the number rather than the actual number itself.
Now things are a little clearer. It seems like there is a relationship between the number of zeros in the spotify plays and how high I rated the song. This make sense, the better the song is, the more people will want to listen to it.
My next question was, where did I and the public, as defined by spotify plays, most disagree. To do this I performed a linear regression, a technique that tries to find a simple line that describes the data as best as possible. I then found how far each of the songs was away from this line. If a song was above this line, I called it underrated, because I thought it was better than the number of listens it had indicated. If it was below the line, I called it overrated, because it was played more times than it should be. The result of this is the following plot.
The purple line is what was recovered. Interesting the line's slope is almost one to one between the score and the number of 0's in the spotify plays. A song with 10,000 listens would theoretically be 0. And a score with 1,000,000,000 listens would be a 5. The exact formula for the line is:
Score = 1.01 * log(popularity) - 4.15
The most overrated song was "When I Kissed the Teacher," which I did not like at all.
The most underrated song was "The Day Before You Came," which I had never heard of before. I did think is was one of the best of the "sad ABBA" songs that start becoming more prevalent near the end of their original run.
Data
You can download all of the data, including the over- and under-ratedness scores by clicking here: Data