Guess Billboard Hot 100?

“Guess Billboard Hot 100?” is a ‘FEAT.’ platform to choose two artists, “who ‘FEAT.’ whom”
and provide the expected rank on Billboard Hot 100 Chart based on machine learning technology.



SINCE the social network has taken up the center stage in every aspect of our life, the indicator of the popularity in the entertainment business has been shifted. Especially in the music scene, the record sales or the frequent exposure to the major media yields their limelight to the new medium, the social network and youtube. The number of followers of the social network and the subscribers of youtube can be the reliable proof for the popularity of the pop artists? How can these numbers be related with the commercial success of the song? This project started from these questions.

Featuring is the particular type of creative collaboration involving one artist integrating another artist’s contribution, either instrumentally or vocally, into their work and publicizing it with a "featuring" credit. It has become the key element to the music fan ‘who feat. whom’ and played a significant role in the marketing and publicity for the song even before the release.

“Guess Billboard Hot 100?” is an interactive platform to choose two artists, “who ‘FEAT.’ whom” and provide the expected rank on Billboard Hot 100 Chart based on machine learning technology.

What we want to examine with this platform is the plausibility of an artist's popularity on Social networks for the song’s success on the chart rather than the music’s quality itself.

Methodology

1. Data for Featuring

Among the songs of the Billboard Hot 100 Chart from 2010 to 2020, We select the ‘FEATURED’ songs of two artists’. The highest rank of the song and the two artists are the main factors.

2. Data for Artists

The number of the subscribers of Youtube and the total sum of followers of social networks are the basic data for each artist.

3. Model and Training

We combine ‘Data for Featuring’ and ‘Data for Artists’ to generate a dataset. The prediction model is generated by Keras’ Sequential model.



Result

Comparison of the actual rank with the predicted result by the platform on the Billboard Hot 100 Chart in June 2021

MAIN ARTISTS FEAT. ARTISTS SONG ACTUAL PREDICTION
Future Lil Uzi Vert Drankin N Smokin 31 24
A Boogie Wit Da Hoodie Lil Durk 24 Hours 88 94
J. Cole Bas 1 0 0 . m i l’ 14 28
Machine Gun Kelly blackbear my ex’s best friend 20 2
J. Cole Lil Baby p r i d e . i s . t h e . d e v i l 7 29
Maroon 5 Megan Thee Stallion Beautiful Mistakes 13 24
H.E.R. Chris Brown Come Through 64 45
Polo G G Herbo Go Part 1 86 65
Polo G DaBaby Party Lyfe 85 85
Migos Polo G Malibu 65 54

Conclusion

By testing several pairs of artists on this platform, we can provide the hypothetical ‘Feat’ pair and predict its rank on the chart. We also can see that it is possible to predict the song’s hit on the chart only by the indicators of the popularity on social networks. Even in the same pair of the artists, the vice-versa of ‘WHO’ feat ‘Whom’ leads to the different outcome on the chart. While establishing the data set, the numbers of the followers and subscribers are steadily changed(mostly added). A consistent check-up is needed to keep it up.

For further development, other factors besides the artists’ popularity, like the artists’ creative capacity(the composing or producing skill), the budget for the production and promotion will be added for more prediction.

Images of artists used on this platform are taken from Wikimedia Commons.