Prediksi Tren Lagu Indonesia di Masa Pandemi Menggunakan Model Recurrent Neural Network

Joedhiawan, Muhammad Akmal (2023) Prediksi Tren Lagu Indonesia di Masa Pandemi Menggunakan Model Recurrent Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05111940000125-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (4MB) | Request a copy

Abstract

Musik menjadi media hiburan bagi banyak orang sejak dulu hingga saat ini. Perkembangan musik ditandai dengan maraknya konser musik di seluruh penjuru tanah air. Namun, pandemi merubah cara masyarakat dalam mendengarkan musik menjadi menggunakan aplikasi. Spotify menjadi salah satu aplikasi musik paling sering digunakan di Indonesia dimana perkembangan Spotify selama pandemi meningkat tajam, Namun, kondisi ini membuat tren popularitas musik pada Spotify menjadi kurang jelas ketika pandemi. Pada tugas akhir ini dilakukan prediksi mengenai tren musik Indonesia pada aplikasi Spotify berdasarkan kemiripan fitur audio untuk memprediksi tren popularitas saat pandemi dan melakukan analisis tren popularitas musik sebelum dan saat pandemi. Prediksi pada tugas akhir ini menggunakan deep learning yaitu recurrent neural network (RNN), long short-term memory (LSTM), dan Gated Recurrent Unit (GRU). Pada Algoritma GRU mendapat MAE sebesar 0,0641 dan RMSE sebesar 0,0779. Model kedua ditempati oleh model LSTM dengan MAE sebesar 0,0772 dan RMSE sebesar 0,0907. Model terakhir adalah RNN dengan MAE 0,0921 sebesar dan RMSE sebesar 0,1100. Hal ini juga terlihat dalam pengujian akurasi untuk prediksi kategori tertinggi bulanan dengan LSTM dan GRU sama-sama mendapat akurasi tertinggi dengan 0,6815 dan RNN memperoleh 0,4425. Algoritma GRU mendapatkan nilai error terendah dan tingkat akurasi tertinggi dengan kategori musik 2 memiliki tren popularitas prediksi paling tinggi daripada kategori musik lain sebelum dan saat pandemi. Kategori musik 2 memiliki nilai acousticness paling tinggi daripada kategori musik 1 dan kategori musik 3 serta fitur energy, liveness, dan valence dengan nilai fitur audio paling rendah daripada kategori musik 1 dan kategori musik 3.
========================================================================================================================
Music has been a medium of entertainment for many people since ancient times until now. The development of music is marked by the rise of music concerts throughout the country. However, the pandemic changed the way people listen to music to use applications. Spotify is one of the most used music applications in Indonesia where Spotify's development during the pandemic has increased sharply. However, this condition has made the trend of music popularity on Spotify less clear during the pandemic. In this final project, predictions are made regarding Indonesian music trends on the Spotify application based on the similarity of audio features to predict popularity trends during a pandemic and to analyze music popularity trends before and during the pandemic. The predictions in this final project use deep learning, namely recurrent neural network (RNN), long short-term memory (LSTM), and Gated Recurrent Unit (GRU). The GRU Algorithm gets an MAE of 0.0641 and an RMSE of 0.0779. The second model is occupied by the LSTM model with an MAE of 0.0772 and an RMSE of 0.0907. The last model is the RNN with an MAE of 0.0921 and an RMSE of 0.1100. This can also be seen in the assessment test for the prediction of the highest monthly category with LSTM and GRU both getting the highest score with 0.6815 and RNN getting 0.4425. The GRU algorithm gets the lowest error value and the highest level of accuracy with music category 2 having the highest trend prediction popularity of other music categories before and during the pandemic. Music category 2 has the highest acoustic value than music category 1 and music category 3 and features energy, liveness, and valence with the lowest audio feature value than music category 1 and music category 3.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deep Learning, Pandemi, Popularitas Musik, Spotify, Tren Popularitas
Subjects:	Q Science Q Science > Q Science (General) Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Akmal Joedhiawan
Date Deposited:	31 Jul 2023 04:02
Last Modified:	31 Jul 2023 06:25
URI:	http://repository.its.ac.id/id/eprint/101089

Actions (login required)

View Item