PRASETYO, Kurniawan Willy and Warsito, Budi and Surarso, Bayu (2025) HIGH-ORDER HESITANT FUZZY TIME SERIES-MULTI LAYER PERCEPTRON DENGAN MEAN AGGREGATED MEMBERSHIP VALUE UNTUK MEMPREDIKSI POLUSI UDARA. Masters thesis, UNIVERSITAS DIPONEGORO.

	Text COVER-1 (1).pdf Download (61kB)
	Text COVER.pdf Restricted to Repository staff only Download (1MB)
	Text BAB I.pdf Download (321kB)
	Text BAB II.pdf Download (716kB)
	Text BAB III.pdf Restricted to Repository staff only Download (888kB)
	Text BAB IV.pdf Restricted to Repository staff only Download (1MB)
	Text BAB V.pdf Restricted to Repository staff only Download (275kB)
	Text DAFTAR PUSTAKA.pdf Download (244kB)
	Text LAMPIRAN.pdf Restricted to Repository staff only Download (381kB)

Abstract

Polusi udara merupakan salah satu permasalahann lingkungan yang dapat berdampak langsung bagi kehidupan manusia. Penelitian ini mencoba untuk mengembangkan model prediksi dengan mengintegrasikan high order hesitant fuzzy time series, multilayer perceptron, mean aggregated membership value, dan length-based discretization. HOHFTS digunakan untuk menangkap ketidakpastian dan dinamika data deret waktu secara lebih tajam melalui struktur fuzzy tingkat lanjut. MLP berperan dalam memodelkan hubungan nonlinear antar data melalui FLR tanpa proses defuzzifikasi, sementara mean aggregated membership value menyederhanakan kompleksitas fuzzy. LBD memungkinkan pembentukan interval partisi secara otomatis dan adaptif terhadap variasi data. Model ini tidak memerlukan asumsi statistik formal seperti normalitas dan stasioneritas, sehingga fleksibel terhadap data lingkungan yang kompleks. Data yang digunakan dalam penelitian ini merupakan data konsentrasi 7 parameter polutan di Kota Semarang dari 1 Januari 2023 sampai 4 September 2024. Hasil menunjukkan model HOHFTS-MLP dapat memprediksi data dengan sangat akurat yang ditunjukkan oleh tingkat akurasi SMAPE kurang dari 10% dalam memprediksi SO_2,O_3,HC,NO_2,PM_10 dan PM_2.5. Model yang dikembangkan diimplementasikan dalam sebuah sistem informasi yang dapat menampilkan prediksi konsentrasi polutan dan klasifikasi tingkat polusi secara interaktif.
Kata Kunci: High Oder Fuzzy Time Series, Multilayer Perceptron, Hesitant Fuzzy, Prediksi polusi udara, Metrik Akurasi

Air pollution is one of the major environmental issues that directly affects human life. This study aims to develop a predictive model by integrating high order hesitant fuzzy time series, multilayer perceptron, mean aggregated membership value, and length-based discretization. The high order hesitant fuzzy approach captures uncertainty and time series dynamics more effectively through advanced fuzzy structures. The multilayer perceptron models nonlinear relationships among data within the fuzzy logical relationship without requiring a defuzzification process, while the mean aggregated membership value simplifies the complexity of the fuzzy system. The length-based discretization enables automatic and adaptive partitioning of intervals based on data variation. The model does not rely on formal statistical assumptions such as normality and stationarity, making it more flexible for handling complex environmental data. The dataset used in this study includes the concentration of seven air pollutant parameters in Semarang City from January 1st, 2023 to September 4th, 2024. Results show that the proposed model can predict pollutant concentrations with high accuracy, indicated by SMAPE values of less than 10% for SO_2,O_3,HC,NO_2,PM_10 and PM_2.5. The developed model has been implemented in an information system that provides interactive air pollutant concentration predictions and pollution level classifications.
Keywords: High Oder Fuzzy Time Series, Multilayer Perceptron, Hesitant Fuzzy, Air Pollution Prediction, Accuracy Metric

Actions (login required)

View Item