Master Thesis Online
If we had to do it again - an algorithmic view of the magic formula behind a commercially successful French hip-hop song
Zolotariov, Denis 
Promotor(s) :
Ittoo, Ashwin
Date of defense : 4-Sep-2023/8-Sep-2023 • Permalink : http://hdl.handle.net/2268.2/18786
ZOLOTARIOV_Denis___Master_Thesis.pdfDetails
| Title : | If we had to do it again - an algorithmic view of the magic formula behind a commercially successful French hip-hop song |
| Translated title : | [fr] "Si c'était à refaire" - une vue algorithmique sur la formule magique derrière le succès commercial d'une chanson hip-hop française |
| Author : | Zolotariov, Denis
|
| Date of defense : | 4-Sep-2023/8-Sep-2023 |
| Advisor(s) : | Ittoo, Ashwin
|
| Committee's member(s) : | Chuor, Porchourng
|
| Language : | English |
| Number of pages : | 99 |
| Keywords : | [en] music [en] topic modelling [en] hec liège [en] digital business [en] model training [en] prediction model [en] latent dirichlet allocation [en] music certification |
| Discipline(s) : | Business & economic sciences > Multidisciplinary, general & others |
| Target public : | Professionals of domain Student |
| Institution(s) : | Université de Liège, Liège, Belgique |
| Degree: | Master en ingénieur de gestion, à finalité spécialisée en digital business |
| Faculty: | Master thesis of the HEC-Ecole de gestion de l'Université de Liège |
Abstract
[en] The aim of this research thesis is trying to define what features of a French hip-hop song contribute to its commercial success. To do so, we use different online sources to work: the French national music certification organisation, called SNEP, serves us for retrieving data on certified albums; Genius, the leader website for music lyrics is used to extract the lyrics content of each song; LyricsGenius, a Python API for Genius, ensures the link between Genius and our Python code. After the data collection step, we end up creating a database storage system, hosted in MongoDB. Re-using this database with Python's library Pandas, we then train an oversampled version of the Random Forest algorithm (after extensive trials on different prediction algorithms) to reach a 81% accuracy in our F1-score. However, assumptions were made: mainly, we focused on the certified albums to analyse the success of songs. Such things are discussed and nuanced in limitations, and with the addition to our scientific literature review, help us define the future research paths that look very promising.
File(s)
Document(s)
Cite this master thesis
The University of Liège does not guarantee the scientific quality of these students' works or the accuracy of all the information they contain.