PERBANDINGAN OPTIMASI PSO DAN RANDOM SEARCH PADA ALGORITMA LIGHTGBM UNTUK KLASIFIKASI TINGKAT PENCEMARAN UDARA

Alifiansyah, Muchamad Dicky (2025) PERBANDINGAN OPTIMASI PSO DAN RANDOM SEARCH PADA ALGORITMA LIGHTGBM UNTUK KLASIFIKASI TINGKAT PENCEMARAN UDARA. Undergraduate thesis, UPN "VETERAN" JAWA TIMUR.

	Text (Cover) 21081010276_Cover.pdf Download (815kB)
	Text (BAB 1) 21081010276_bab 1.pdf Download (260kB)
	Text (BAB 2) 21081010276_bab 2.pdf Restricted to Repository staff only until 2027. Download (395kB)
	Text (BAB 3) 21081010276_bab 3.pdf Restricted to Repository staff only until 2027. Download (403kB)
	Text (BAB 4) 21081010276_bab 4.pdf Restricted to Repository staff only until 2027. Download (978kB)
	Text (BAB 5) 21081010276_bab 5.pdf Download (191kB)
	Text (DAFTAR PUSTAKA) 21081010276_dafpus.pdf Download (192kB)
	Text (LAMPIRAN) 21081010276_lampiran.pdf Restricted to Repository staff only Download (265kB)

Abstract

Clean, high-quality air plays a role in affecting the health of living things. However, in reality, the air is not completely clean due to pollution that causes air pollution. The effects of air pollution can cause diseases such as respiratory problems, asthma, lung cancer, tuberculosis, heart disease, respiratory tract infections, and eye diseases. Therefore, it is important to monitor air pollution in order to plan mitigation and action in the event of air pollution. Manual monitoring of air pollution with the naked eye and passive samplers has limitations in terms of efficiency, practicality, and time. The application of Machine Learning can be a solution in monitoring and monitoring air pollution based on data. This study aims to build an air pollution level classification model using the LightGBM algorithm, which will be combined with optimization techniques. The PSO and Random Search optimization techniques play a role in improving the performance of LightGBM in classification. Random Oversampling is used to handle data imbalance and uses two data ratios, namely 80:20 and 70:30. The results show that the LightGBM model with Random Search and Oversampling and an 80:20 ratio produced the highest accuracy of 0.9966 and an execution time of 79.00 seconds. Meanwhile, LightGBM with PSO and Oversampling and a ratio of 80:20 had the same accuracy, but the execution time was much longer at 235.18 seconds. Random Search had better results than PSO in improving model performance, as evidenced by the fact that the model without Oversampling Random Search was able to outperform PSO with an accuracy of 0.9755. Models combined with Oversampling have better accuracy results than models without Oversampling, meaning that Oversampling has an influence on improving model performance. Overall, models with Random Search optimization have better performance and results than models with PSO optimization, both with and without Oversampling

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Rahajoe, Ani Dijah	NIDN0012057301	anidijah.if@upnjatim.ac.id
Thesis advisor	Puspaningrum, Eva Yulia	NIDN0005078908	evapuspaningrum.if@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Muchamad Dicky Alifiansyah

Date Deposited:

05 Dec 2025 08:51

Last Modified:

05 Dec 2025 08:51

URI:

https://repository.upnjatim.ac.id/id/eprint/48074