IMPLEMENTASI ALGORITMA RANDOM FOREST UNTUK KLASIFIKASI INDEKS STANDAR PENCEMARAN UDARA (ISPU) DI PROVINSI DKI JAKARTA

ELANG PUTRA KARUNIAWAN, . (2026) IMPLEMENTASI ALGORITMA RANDOM FOREST UNTUK KLASIFIKASI INDEKS STANDAR PENCEMARAN UDARA (ISPU) DI PROVINSI DKI JAKARTA. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text Cover.pdf Download (1MB)
	Text BAB I.pdf Download (550kB)
	Text BAB II.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB III.pdf Restricted to Registered users only Download (792kB) | Request a copy
	Text BAB IV.pdf Restricted to Registered users only Download (846kB) | Request a copy
	Text BAB V.pdf Restricted to Registered users only Download (340kB) | Request a copy
	Text Daftar Pustaka.pdf Download (333kB)
	Text Lampiran.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Pencemaran udara di wilayah perkotaan, khususnya Provinsi DKI Jakarta, memerlukan sistem klasifikasi kualitas udara yang mampu menganalisis hubungan kompleks antar parameter polutan secara lebih adaptif dibandingkan pendekatan konvensional. Penelitian ini bertujuan membangun model klasifikasi kualitas udara berbasis machine learning menggunakan algoritma Random Forest. Dataset yang digunakan adalah data historis Indeks Standar Pencemaran Udara (ISPU) DKI Jakarta periode Januari 2024–November 2025. Tahapan penelitian meliputi data cleaning, data transformation, pembagian data latih dan data uji, optimasi hyperparameter menggunakan GridSearchCV, serta pemodelan Random Forest melalui dua skenario, yaitu baseline dan optimasi menggunakan SMOTE. Evaluasi model dilakukan menggunakan Confusion Matrix dengan metrik accuracy, precision, recall, dan F1-score. Hasil pengujian menunjukkan kedua skenario menghasilkan nilai akurasi global yang stabil sebesar 0.9871 (99%). Namun, penerapan SMOTE berhasil meningkatkan sensitivitas model pada kelas minoritas, ditunjukkan dengan melonjaknya nilai recall kategori "Baik" dari 0.91 menjadi 0.94, serta memperkecil celah selisih (gap accuracy) latih-uji menjadi sangat ideal di angka 0.49%. Selain itu, analisis Feature Importance mencatatkan parameter partikulat halus PM2.5 sebagai fitur paling dominan dengan bobot kontribusi mencapai 67.2%. Dapat disimpulkan bahwa kombinasi Random Forest mampu menghasilkan model klasifikasi ISPU yang akurat, stabil, dan memiliki kemampuan generalisasi yang baik, sedangkan penerapan SMOTE efektif dalam meningkatkan keseimbangan kemampuan klasifikasi terhadap kelas minoritas tanpa mengurangi performa model secara keseluruhan. ***** Air pollution in urban areas, particularly in the Special Capital Region of Jakarta, requires an air quality classification system capable of analyzing the complex relationships among pollutant parameters in a more adaptive manner than conventional approaches. This study aims to develop a machine learning-based air quality classification model using the Random Forest algorithm. The dataset used consists of historical Air Pollution Standard Index (ISPU) data for DKI Jakarta from January 2024 to November 2025. The research stages included data cleaning, data transformation, splitting the data into training and test sets, hyperparameter optimization using GridSearchCV, and Random Forest modeling through two scenarios: a baseline and an optimization using SMOTE. Model evaluation was performed using a Confusion Matrix with the metrics accuracy, precision, recall, and F1-score. The test results show that both scenarios produced a stable global accuracy of 0.9871 (99%). However, the application of SMOTE successfully improved the model’s sensitivity for the minority class, as evidenced by a surge in the recall value for the “Good” category from 0.91 to 0.94, as well as a reduction in the training-test accuracy gap to an ideal 0.49%. Furthermore, the Feature Importance analysis identified the PM2.5 fine particulate matter parameter as the most dominant feature, with a contribution weight of 67.2%. It can be concluded that the Random Forest combination is capable of producing an accurate, stable, and well-generalizing ISPU classification model, while the application of SMOTE is effective in improving the balance of classification performance for the minority class without compromising the model’s overall performance.

Item Type:	Thesis (Sarjana)
Additional Information:	1). Dr. Widodo, S.Kom., M.Kom. ; 2). Neng Ayu Herawati, S.Pd., M.T.
Subjects:	Teknologi dan Ilmu Terapan > Teknik Komputer
Divisions:	FT > S1 Pendidikan Teknik Informatika Komputer
Depositing User:	Elang Putra Karuniawan .
Date Deposited:	10 Aug 2026 02:56
Last Modified:	10 Aug 2026 02:56
URI:	http://repository.unj.ac.id/id/eprint/68754

Actions (login required)

View Item