ANALISIS PERBANDINGAN ALGORITMA LINEAR REGRESSION, RANDOM FOREST, DAN XGBOOST UNTUK MEMPREDIKSI ANGKA PARTISIPASI SEKOLAH (APS) DI INDONESIA

ADRIE BAGAS SAPUTRA, . (2025) ANALISIS PERBANDINGAN ALGORITMA LINEAR REGRESSION, RANDOM FOREST, DAN XGBOOST UNTUK MEMPREDIKSI ANGKA PARTISIPASI SEKOLAH (APS) DI INDONESIA. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text COVER.pdf Download (662kB)
	Text BAB 1.pdf Download (146kB)
	Text BAB 2.pdf Restricted to Registered users only Download (407kB) | Request a copy
	Text BAB 3.pdf Restricted to Registered users only Download (635kB) | Request a copy
	Text BAB 4.pdf Restricted to Registered users only Download (881kB) | Request a copy
	Text BAB 5.pdf Restricted to Registered users only Download (142kB) | Request a copy
	Text DAFTAR PUSTAKA.pdf Download (214kB)
	Text LAMPIRAN & DAFTAR RIWAYAT HIDUP.pdf Restricted to Registered users only Download (2MB) | Request a copy

Abstract

Angka Partisipasi Sekolah (APS) merupakan indikator penting untuk mengukur keberhasilan sistem pendidikan di suatu wilayah. Penelitian ini bertujuan untuk membandingkan kinerja tiga algoritma Machine Learning, yaitu Linear Regression, Random Forest, dan XGBoost, dalam memprediksi nilai APS di 34 provinsi di Indonesia berdasarkan data dari tahun 2003 hingga 2024. Penelitian dilakukan melalui tahapan data preparation, data preprocessing, pembangunan model, serta evaluasi menggunakan lima metrik utama: Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Mean Square Error (MSE), Root Mean Square Error (RMSE), dan R-squared (R²). Seluruh model juga dioptimasi melalui teknik Hyperparameter Tuning menggunakan GridSearchCV dan RandomSearchCV. Hasil penelitian menunjukkan bahwa algoritma XGBoost dengan RandomSearchCV memberikan performa terbaik, dengan nilai MAE sebesar 0.19, MAPE 19.46%, MSE 0.10, RMSE 0.32, dan R² sebesar 0.91. Hal ini menunjukkan bahwa model tersebut mampu menghasilkan prediksi APS dengan kesalahan yang rendah dan akurasi tinggi. Sebaliknya, Linear Regression menunjukkan kinerja yang paling rendah, dan tuning pada Random Forest justru menurunkan performa model. Penelitian ini memberikan kontribusi dalam pemanfaatan algoritma Machine Learning untuk mendukung pengambilan kebijakan pendidikan berbasis data, khususnya dalam upaya meningkatkan partisipasi sekolah di Indonesia. ***** The School Participation Rate (SPR) is an important indicator for measuring the success of an education system in a region. This study aims to compare the performance of three Machine Learning algorithms, namely Linear Regression, Random Forest, and XGBoost, in predicting SPR values in 34 provinces in Indonesia based on data from 2003 to 2024. The research was conducted through the following stages: data preparation, data preprocessing, model building, and evaluation using five main metrics: Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Mean Square Error (MSE), Root Mean Square Error (RMSE), and R-squared (R²). All models were also optimized through Hyperparameter Tuning techniques using GridSearchCV and RandomSearchCV. The results showed that the XGBoost algorithm with RandomSearchCV performed the best, with MAE of 0.19, MAPE of 19.46%, MSE of 0.10, RMSE of 0.32, and R² of 0.91. This indicates that the model is capable of producing APS predictions with low error and high accuracy. Conversely, Linear Regression showed the lowest performance, and tuning on Random Forest actually reduced the model's performance. This study contributes to the utilization of Machine Learning algorithms to support data-driven educational policy-making, particularly in efforts to increase school participation in Indonesia.

Item Type:	Thesis (Sarjana)
Additional Information:	1). Murien Nugraheni, ST., M.Cs. ; 2). Ali Idrus, S.Kom., M.Kom.
Subjects:	Teknologi dan Ilmu Terapan > Teknik Komputer
Divisions:	FT > S1 Sistem dan Teknologi Informasi
Depositing User:	Adrie Bagas Saputra .
Date Deposited:	07 Aug 2025 03:03
Last Modified:	07 Aug 2025 03:03
URI:	http://repository.unj.ac.id/id/eprint/58128

Actions (login required)

View Item