Analisis Terhadap Prediksi Timbulan Sampah di Indonesia Menggunakan Gradient Boosting Regression Tree (GBRT) dan Shapley Additive Explanation (SHAP)

Febriana, Nadira Hanindhita (2026) Analisis Terhadap Prediksi Timbulan Sampah di Indonesia Menggunakan Gradient Boosting Regression Tree (GBRT) dan Shapley Additive Explanation (SHAP). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002211092-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Sampah masih menjadi permasalahan yang dihadapi Indonesia hingga saat ini. Per 21 Agustus 2025, timbulan sampah Indonesia pada tahun 2024 berdasarkan Sistem Informasi Pengelolaan Sampah Nasional (SIPSN) mencapai 34,63 juta Ton dengan 59,92% nya belum terkelola. Sampah-sampah tersebut dapat memberikan dampak buruk bagi lingkungan dan kesehatan masyarakat. Timbulan sampah sendiri berkaitan dengan faktor dari berbagai aspek, seperti sosio-demografi, lingkungan, maupun ekonomi. Maka dari itu, pada penelitian ini digunakan pendekatan machine learning, yaitu Gradient Boosting Regression Tree (GBRT) untuk memprediksi timbulan sampah tingkat Kabupaten/Kota di Indonesia dari data yang kompleks serta digunakan metode Shapley Additive Expalanation (SHAP) untuk menginterpretasi hasil yang diperoleh dari model agar dapat menambah pemahaman mengenai permasalahan sampah. Pada penelitian ini, fitur yang digunakan meliputi jumlah penduduk, kepadatan penduduk, rata-rata lama sekolah, jumlah curah hujan, temperatur rata-rata, Produk Domestik Regional Bruto Atas Dasar Harga Konstan (PDRB ADHK), dan pengeluaran per kapita disesuaikan. Model GBRT yang dibangun menunjukkan performansi baik dengan R^2 sebesar 0,96; RMSE sebesar 33.609,02 Ton; dan MAE sebesar 18.043,96 Ton. Berdasarkan interpretasi SHAP, didapatkan bahwa Jumlah Penduduk, Kepadatan Penduduk, Rata-Rata Lama Sekolah, dan PDRB ADHK merupakan empat fitur dengan kontribusi tertinggi terhadap hasil prediksi. Berdasarkan interpretasi global, nilai SHAP yang dihasilkan menunjukkan bahwa nilai fitur Jumlah Penduduk, Kepadatan Penduduk, Rata-Rata Lama Sekolah, dan PDRB ADHK yang tinggi cenderung berkontribusi meningkatkan hasil prediksi.
==================================================================================================================================
Waste remains a problem faced by Indonesia to date. As of 21 August 2025, waste generation in Indonesian in 2024 reached a total of 34.63 million tons, with 59.92$ not subject to proper management as reported on National Waste Management Information System (SIPSN). These umanaged properly waste have adverse impact on ecosystems and public health. Waste generation itself is related to many factors from multiple aspects, such sociodemographic, environmental, and economic. Hence in this study, machine learning based approach, Gradient Boosting Regression Tree (GBRT) is employed to predict waste generation at the regency and city level from complex data as well as implementing SHAP for model interpretation to provide a deeper understanding of waste-related issues. The features considered in this study include population, population density, average years of education, annual rainfall, average annual temperature, Gross Domestic Regional Product (GDRP) at Constant Price, and expenditure per capita. The GBRT model shows good performance with R2 = 0.96, an RMSE of 33,609.02 Tons, and an MAE of 18,043.96 Tons. Based on interpretion derived from SHAP, it was found that population, population density, average years of education, and GDRP at Constant Price is the four features with the highest contribution to the prediction results. Based on global interpretation, the obtained SHAP values indicate that the higher value of population, population density, average years of education, and GDRP at Constant Price corresponds to an increase in the predicted outcome.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	gradient boosting regression tree, prediksi, shapley additive explanation, timbulan sampah, gradient boosting regression tree, prediction, shapley additive explanation, waste generation
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA336 Artificial Intelligence
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Nadira Hanindhita Febriana
Date Deposited:	28 Jan 2026 08:10
Last Modified:	28 Jan 2026 08:10
URI:	http://repository.its.ac.id/id/eprint/130818

Actions (login required)

View Item