In this graph, we see the frequency of the sentiment score of the songs overall. We have examined the lyrics of the songs and output a sentiment value that lies between -1 and 1. -1 means that the song was very sad, whereas a score of 1 or a value close to 1 means that the song is associated with happy emotions. What is clear to see is that there are hardly any songs that have a value that does not fall into the maximum or minimum values. So there are mainly songs which are associated with negative emotions or most which are associated with very positive emotions.
When analyzing the top 20 songs over the past years, a clear negative trend emerged despite some fluctuations. This result fits well with Spotify's own value of valence. In this evaluation, though, only the lyrics were examined. The sentiment value returns a similar statement as the valence value. Here, too, a higher value means more positive emotions than a negative value.
In the analysis, it must be emphasized that it is a relatively small starting point of data, as we could only use the top 20 songs per year due to the length of the lyrics. In addition, outliers have a very strong influence on the value. Overall, however, a trend can be seen that shows that more and more different words appear in the songs. It is interesting, however, that the number of words increases somewhat towards the beginning of the 2000s, which can be explained by the rap high. A rap song has on average far more words than a non-rap song.
This graph shows the evolution of keywords in popular music over the last decades. The bars represent the most common words used in songs during the decades from the 50s to the 2020s. One realisation is that love is simply not to be thought out of music - no matter where you look, love is everywhere. Somehow the word 'feat' slipped through the pre-processing, but has an interesting insight as a result: the trend of featuring other artists started around 2000 and kept growing since then.