Evaluasi Regresi Terklaster Fuzzy Spasial Simultan dengan Pendekatan Simulasi
DOI:
https://doi.org/10.12962/limits.v22i3.5425Kata Kunci:
regresi terklaster simultan, klasterisasi spasial, spatial fuzzy, SFCR, FGWCRAbstrak
Data spasial merupakan data yang memuat informasi yang berkaitan dengan karakteristik geografis suatu wilayah. Perkembangan data spasial yang mengarah pada data berskala besar membutuhkan metode analisis yang efisien dalam proses pengolahannya. Salah satu metode analisis yang dapat digunakan untuk mengolah data spasial berskala besar adalah spatial fuzzy clustering. Metode ini memungkinkan adanya penyesuaian bobot kelompok berdasarkan kemungkinan data, sehingga lebih mampu menangkap variasi lokal yang sebenarnya terjadi dalam data spasial. Metode spatial fuzzy clustering dengan penalti spasial, Spatial Fuzzy Clustered Regression (SFCR) dan tanpa penalti spasial, Fuzzy Geographically Weighted Clustering Regression (FGWCR) dievaluasi melalui simulasi pada penelitian ini. SFCR merupakan metode yang menggabungkan klasterisasi spasial dan pembentukan persamaan regresi secara simultan, sehingga waktu komputasi menjadi lebih efisien. FGWCR menghasilkan klaster yang mempertimbangkan kedekatan spasial dan kesamaan atribut sehingga efektif digunakan pada data spasial. Data dirancang sehingga terdapat 6 klaster dalam proses simulasi. Hasil simulasi menunjukkan metode SFCR lebih mampu mencerminkan keragaman data dan pembagian klaster dengan akurat. Nilai untuk metode SFCR pada derajat fuzziness 2 dan autokorelasi spasial lemah, sedang, dan kuat berturut-turut yaitu 99.7%, 99.6%, dan 99.5%, sedangkan untuk metode FGWCR yaitu 98.5%, 98.6%, dan 98.1%. Kebaikan persamaan dievaluasi oleh nilai RMSE. Semakin kecil nilai RMSE maka persamaan yang dihasilkan semakin baik. Nilai RMSE untuk metode SFCR pada derajat fuzziness 2 dan autokorelasi spasial lemah, sedang, dan kuat berturut-turut yaitu 0.30, 0.289, dan 0.298, sedangkan untuk metode FGWCR yaitu 0.659, 0.541, dan 0.551.
Spatial data refers to data that contains information related to the geographical characteristics of a region. As spatial data evolves into large-scale datasets, efficient analytical methods are required for processing the data. One such method suitable for analyzing large-scale spatial data is spatial fuzzy clustering. This method allows for the adjustment of cluster weights based on data likelihood, making it more capable of capturing the actual local variations present in spatial data. In this study, two types of spatial fuzzy clustering methods were evaluated through simulation: the method with a spatial penalty, Spatial Fuzzy Clustered Regression (SFCR), and the method without a spatial penalty, Fuzzy Geographically Weighted Clustering Regression (FGWCR). SFCR is a method that combines spatial clustering and regression modeling simultaneously, resulting in more efficient computation time. FGWCR produces clusters by considering both spatial proximity and attribute similarity, making it effective for spatial data analysis. The data were designed to form six clusters during the simulation process. The simulation results showed that the SFCR method was more capable of accurately capturing data variation and cluster distribution. The R² values for SFCR at a fuzziness degree of 2 and under weak, moderate, and strong spatial autocorrelation were 99.7%, 99.6%, and 99.5%, respectively, while the R² values for FGWCR were 98.5%, 98.6%, and 98.1%. Model performance was evaluated using RMSE, where lower RMSE values indicate better performance. The RMSE values for the SFCR method at a fuzziness degree of 2 and under weak, moderate, and strong spatial autocorrelation were 0.30, 0.289, and 0.298, respectively, while the RMSE values for the FGWCR method were 0.659, 0.541, and 0.551.
Referensi
A. Djuraidah, Monograph Penerapan dan Pengembangan Regresi Spasial dengan Studi Kasus pada Kesehatan, Sosial, dan Ekonomi. Bogor: IPB Press, 2020.
S. Cho, D. M. Lambert, S. G. Kim, dan S. Jung, “Extreme coefficients in geographically weighted regression and their effects on mapping,” GIScience Remote Sens., vol. 46, no. 3, hal. 273–288, 2009, doi: 10.2747/1548-1603.46.3.273.
S. Sugasawa dan D. Murakami, “Spatially clustered regression,” Spat. Stat., vol. 44, hal. 100525, Agu 2021, doi: 10.1016/j.spasta.2021.100525.
V. K. Malhotra, H. Kaur, dan M. A. Alam, “An Analysis of Fuzzy Clustering Methods,” Int. J. Comput. Appl., vol. 94, no. 19, hal. 9–12, 2014.
A. W. Wijayanto, “Perbaikan Algoritma Fuzzy Geo-Demographic Clustering Menggunakan Optimisasi Metaheuristic Pada Data Sensus Penduduk Indonesia,” Institut Teknologi Bandung, 2015.
W. Andrytiarandy, “Analisis Regresi Terboboti Fuzzy Cluster Geografis Pada Prevalensi Malaria Di Indonesia,” Institut Pertanian Bogor, 2017.
M. N. Aidi et al., “Province clustering based on the percentage of communicable disease using the BCBimax biclustering algorithm,” Geospat. Health, vol. 18, no. 2, Sep 2023, doi: 10.4081/gh.2023.1202.
D. Nicholson, O. A. Vanli, S. Jung, dan E. E. Ozguven, “A spatial regression and clustering method for developing place-specific social vulnerability indices using census and social media data,” Int. J. Disaster Risk Reduct., vol. 38, hal. 101224, Agu 2019, doi: 10.1016/j.ijdrr.2019.101224.
F. Li dan H. Sang, “Spatial Homogeneity Pursuit of Regression Coefficients for Large Datasets,” J. Am. Stat. Assoc., vol. 114, no. 527, hal. 1050–1062, Jul 2019, doi: 10.1080/01621459.2018.1529595.
I. K. Hasan, Nurwan, Nur Falaq, dan Muhammad Rezky Friesta Payu, “Optimization Fuzzy Geographically Weighted Clustering with Gravitational Search Algorithm for Factors Analysis Associated with Stunting,” J. RESTI (Rekayasa Sist. dan Teknol. Informasi), vol. 7, no. 1, hal. 120–128, Feb 2023, doi: 10.29207/resti.v7i1.4508.
G. . Mason dan R. . Jacobson, “Fuzzy Geographically Weighted Clustering,” in Proceedings of the 9th International Conference on Geocomputation, 2007, hal. 1–7.
K. Zhou, C. Fu, dan S. Yang, “Fuzziness parameter selection in fuzzy c-means: The perspective of cluster validation,” Sci. China Inf. Sci., vol. 57, no. 11, hal. 1–8, Nov 2014, doi: 10.1007/s11432-014-5146-0.
Fotheringham AS, C. Brunsdon, dan C. M, Geographically Weighted Regression the Analysis of Spatially Varying Relationships. New York: Wiley, 2002.
A. A. Suryanto, “Penerapan Metode Mean Absolute Error (Mea) dalam Algoritma Regresi Linear untuk Prediksi Produksi Padi,” SAINTEKBU, vol. 11, no. 1, hal. 78–83, Feb 2019, doi: 10.32764/saintekbu.v11i1.298.
