The Data
The dataset we used for this webpage consists of ~25 albums for each subgenre. The line between eras and
subgenres can blur, so we tried to stick to the most clear-cut albums and artists for each subgenre. To
decide on the most influential albums for each subgenre, we took various inputs from the internet -
ChatGPT,
Reddit, Wikipedia, and various other websites. Sometimes, albums belong to more than one subgenre. For
example, Straight Outta Compton is classified as both golden age and gangsta rap. Some artists are
purposefully excluded from the dataset because of the diversity of the songs in each of their albums.
For
example, any album by Jay-Z is hard to pin as gangsta rap because of the various themes and styles.
Again,
these are only a few of the subgenres of rap, and even the ones we discuss in this article can be split
up
into smaller subgenres (like my personal favorite, g-funk as a subgenre of gangsta rap).
To get the data on each of the albums, we used a python library called lyricsgenius and web scraped the
data
from Genius.com. All of the data visualizations were made with D3.js, a javascript library. Here is some
more detailed information about each of the visualizations…
-
Artist Bubble Chart - The artists that go into these visualizations are based entirely off of
the
albums
that went into the subgenre. The size of each circle is determined by the artist’s monthly listeners
on
Spotify as of April 2023. It is worth noting that the circles are sized relative only to the artists
of
the
same subgenre. So, the size of a trap artist’s circle should not be compared with the size of an old
school
rapper’s circle. The circles are intractable if you click and drag them.
-
Most Common Words - These words were chosen through a technique called TF-IDF. TF-IDF (term
frequency-inverse document frequency) in this context evaluates how important a word is to a
subgenre
relative to the rest of the subgenres. The metric is calculated based on the frequency of a word in
a
subgenre and its frequency in all of the other subgenres. So, common words that exist across all
albums
are
automatically filtered out. Really, these charts show the most common words that are still fairly
unique
to
that subgenre. The words in this chart are also interactable if you click and drag them.
-
Sentiment Gauge - To calculate the sentiment gauge for a subgenre, we used TextBlob, a python
library, to
get a sentiment score for each word from each album in a given subgenre. Then, we averaged those
scores
to
come up with the final sentiment score for the subgenre.
-
Topics Spider Chart - To create this chart, we came up with lists of words likely to
correlate
with each of
the topics. For example, our list of drug-related words includes “alcohol”, “xanax”, “weed”, etc.
The
algorithm goes through all the lyrics of the given subgenre and counts the number of times the
rappers
said
a drug-related word. We then calculate the percentage of their lyrics that are drug-related so that
subgenres with more/less lyrics are scored fairly.
