Πλοήγηση ανά Συγγραφέα "Iliaki, Georgia"

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Α Β Γ Δ Ε Ζ Η Θ Ι Κ Λ Μ Ν Ξ Ο Π Ρ Σ Τ Υ Φ Χ Ψ Ω

Τώρα δείχνει 1 - 1 από 1

Applications of machine learning on Spotify data
(2021-07-29) Iliaki, Georgia; Ηλιάκη, Γεωργία; Athens University of Economics and Business, Department of Management Science and Technology; Louridas, Panagiotis
This study refers to machine learning applications on data scraped from the Spotify API website. It is divided in two sections based on different data provided by the company. The first section the data handled are the musical features of the songs and an effort is made to classify over 2000 songs based on the emotion they convey to the listener using different classification methods such us Neural Networks, Random Forest, LightGBM, XGboost. Also two regression methodologies are used (Neural Network Regressor and Random Forest Regressor) in order to predict the "valence" value of the songs (how happy or not a song is). On the second part of the analysis the structural layers of the songs are used to create 5 different Neural Network model, one for each layer (Sections, Segments, Tatums, Beats and Bars) to figure out how deep the emotion can be traced on a song. On the first part the most effective method appeared to be the Random Forests. On the second part of the study, the results indicated that the emotions of the songs were better identified on the deepest structural levels of the songs, on the segments data set.