KLASIFIKASI TINGKAT KEPARAHAN KECELAKAAN LALU LINTAS BERBASIS CATBOOST PADA DATA YANG TIDAK SEIMBANG MENGGUNAKAN SMOTENC DAN OPTUNA

Ayatillah, Maslahatul Kaunaini (2026) KLASIFIKASI TINGKAT KEPARAHAN KECELAKAAN LALU LINTAS BERBASIS CATBOOST PADA DATA YANG TIDAK SEIMBANG MENGGUNAKAN SMOTENC DAN OPTUNA. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text COVERR.pdf Download (1MB)
	Text BAB 1.pdf Download (33kB)
	Text BAB 2.pdf Restricted to Repository staff only until 19 January 2029. Download (924kB)
	Text BAB 3.pdf Restricted to Repository staff only until 19 January 2029. Download (671kB)
	Text BAB 4.pdf Restricted to Repository staff only until 19 January 2029. Download (2MB)
	Text BAB 5.pdf Download (132kB)
	Text DAFTAR PUSTAKA.pdf Download (142kB)
	Text LAMPIRAN.pdf Restricted to Repository staff only until 19 January 2029. Download (73kB)

Abstract

Traffic accidents represent a complex issue in Surabaya, having a significant impact on public safety and socio-economic loss. A primary challenge in accident severity classification modeling is the phenomenon of data imbalance, where slight injury cases are considerably more dominant than severe or fatal categories. This study aims to classify accident severity into three classes: Class 0 (Slight), Class 1 (Moderate), and Class 2 (Severe) by implementing the CatBoost algorithm. To address the uneven class distribution, the Synthetic Minority Over-sampling Technique for Nominal and Continuous (SMOTENC) was applied, while hyperparameter optimization was conducted using the Optuna framework. Model evaluation utilized the hold-out method with a 60% training and 40% testing data split. Experimental results demonstrate that the integration of SMOTENC and Optuna significantly enhances the model's predictive performance. The CatBoost-SMOTENC-Optuna model achieved superior performance with an accuracy of 97.20%, precision of 95%, recall of 90%, and an F1-Score of 92%. Overall, this research proves that the combination of SMOTENC and Optuna optimization is effective in handling imbalanced data and increasing model sensitivity across all severity classes. These findings are expected to serve as a scientific reference for authorities in formulating more accurate risk mitigation strategies and traffic safety policies.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Sari, Anggraini Puspita	NIDN0716088605	anggraini.puspita.if@upnjatim.ac.id
Thesis advisor	Junaidi, Achmad	NIDN0710117803	achmadjunaidi.if@upnjatim.ac.id

Subjects:

H Social Sciences > HG Finance > HG1709 Data processing
T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Maslahatul Kaunaini Ayatilah

Date Deposited:

23 Jan 2026 08:47

Last Modified:

23 Jan 2026 08:47

URI:

https://repository.upnjatim.ac.id/id/eprint/48797