Looking at Song Genres, Release Year, and Time Signature
The graphs above show time signature vs. year, along with year. vs. genre. If we look at Fig. 1, we can see that certain time signatures have grown in popularity over time, while others might have decreased. The first bias that must be accounted for it the total number of songs per year in the dataset. There are many more recent songs in the dataset compared to older ones. So, every time signature is more likely to have more occurrences in recent years. However, that does not mean no information can be gained.
If we look at time signature “7” (seven beats per bar), we can see that its first occurrence happens around 1975, and has grown in time since then. Music using this time signature mainly corresponds to the Metal and Jazz genres. So, if we look at Fig. 2 above, we can see that Metal songs do not occur in our dataset until around 1980, and the 7 time signature becomes even more popular shortly after that time.
While the results displayed here are very preliminary findings, the graphics shown display that there is a potential for more information to be gained by looking at how time-signature popularities have changed over time. This would best be done by first normalizing the amount of songs/year and then looking at how many fall into each time signature. More significant information would come from using a larger dataset.