In order to be more representative of top music artists, I suggest that instead of simply counting how many times a song by an artist is played (example: one song and only one song by artist X is played, and then artist X is apparently the top artist), there should be an X and Y axis set up determining the top artists.
On one axis, the number of different songs played by the artist is measured. On the other, the average number of plays per song is represented. (Example: I played 10 songs by artist X and I played them each an average of 13 times VS. the current setup of I played 1 song by artist Y 56 times) In this situation, artist X is a more representative sample of an artist who I’ve listened to more consistently and broadly, implying I enjoy a lot of their music and them as an artist. This, I believe, is a more representative way to calculate top artists for the year rather than simply having one of the artist’s songs somehow meaning they are the top artist, as I could dislike all of their other music.