Overview
First let’s have a look at the distributions of all the numeric variables.
Valence looks an interesting place to start as it is the most symmetrical, indicating an even spread of sad –> happy songs.

Genres
So is there an obvious difference in valence across genres?
edm is generally more negative sounding while latin is more positive

Drilling down to artist level
Let’s shine a light on one particular artist. The more data the better, so who are the artists appearing most frequently?

## # A tibble: 10,693 x 2
##    track_artist                  n
##    <chr>                     <int>
##  1 Martin Garrix               161
##  2 Queen                       136
##  3 The Chainsmokers            123
##  4 David Guetta                110
##  5 Don Omar                    102
##  6 Drake                       100
##  7 Dimitri Vegas & Like Mike    93
##  8 Calvin Harris                91
##  9 Hardwell                     84
## 10 Kygo                         83
## # … with 10,683 more rows

Hmmm…
I’m not really familiar with Martin Garrix or most of these artists to be honest, but Queen were one of my favourite bands growing up, so let’s go further with their data.

The distribution of valence for Queen songs is fairly well spread with a slightly positive skew

How can we group these dots?
The spotify data gives us the songs, but what albums are they from? Who wrote them? What was their date of release?

Queen were unusual in that all band members sung lead on a few songs. It would be interesting to explore patterns in valence when we factor in some or all of these attributes.

This isn’t information we have in the spotify data so I scraped it from Wikipedia, then joined the two datasets together. Now we not only have the songs, but we have a load more intersting variables!

Albums
Most albums have quite a range from positive to negative, but A Day at the Races is notable as being on the whole a ‘negative’ sounding album, with only a single song over 0.5 valence.

It’s worth noting at this point that our spotify data is incomplete, with some albums only having a single song.

Taking a closer look at ‘A Day at the Races’
‘You and I’ has the highest valence on the album but is still only just above 0.5

Songwriters
While Mercury and May songwriting leans towards sad/negative, Deacon and Taylor’s are generally more postive/happy

Clearly a disproportionate number of high valence songs were either Deacon’s or Taylor’s

Getting close to the final design
Freddie Mercury’s songs were often sad and reflective.
John Deacon generally wrote upbeat pop songs

Final thoughts
There were other avenues to explore but this was a good point to stop the analysis.

Valence is interesting because it attempts to measure a kind of human emotion conveyed in the music.
Visualising this by looking at the band members themselves allows us to ask questions about the type of person each was, through the positivity/negativity of their songwriting tendencies.