PENANGANAN DATA OUTLIER MENGGUNAAN ROBUST CLUSTER DALAM PEMETAAN DATASET STUNTING
DOI:
https://doi.org/10.54314/jssr.v8i4.4248Abstract
Abstract:The urgency of this research lies in the importance of accurate stunting data mapping to understand the patterns and distribution of this issue within a specific region. However, in data collection and analysis, outliers—values far beyond the normal range—are often present and may affect the validity of the analysis results. Conventional clustering methods, such as K-Means, are highly sensitive to outliers, which may lead to inaccurate and misleading mappings. Therefore, a more robust approach is required to handle outliers, one of which is the robust clustering method. Robust Clustering is an effective technique in dealing with outliers, enabling the identification of homogeneous data groups without being affected by extreme values. This study aims to explore and address the issue of outliers in stunting dataset mapping. Based on the clustering analysis using the K-Means method, three groups of Public Health Centers (Puskesmas) in Asahan Regency were identified based on stunting cases from 2022 to 2024. After outlier detection and handling, the Davies-Bouldin Index decreased from 0.529 to 0.483, indicating improved cluster quality. The final results show that the majority of Public Health Centers fall into Cluster 1 with relatively low to moderate stunting cases, Cluster 2 consists of centers with significantly increasing cases, and Cluster 3 represents those with very high numbers of cases.
Â
Keywords: Stunting; Data Outlier; Robust Clustering; Mapping; Machine Learning
Â
Abstrak: Urgensi penelitian ini terletak pada pemetaan data stunting yang akurat sangat penting untuk memahami pola dan distribusi masalah ini di wilayah tertentu. Akan tetapi, dalam pengumpulan dan analisis data, seringkali terdapat data outlier, nilai yang jauh di luar rentang normal, yang dapat mempengaruhi hasil analisis. Metode pengelompokan konvensional, seperti K-means, sering kali sensitif terhadap outlier, yang dapat menghasilkan pemetaan yang tidak akurat dan menyesatkan. Oleh karena itu, diperlukan pendekatan yang lebih robust untuk menangani data outlier, salah satunya dengan menggunakan metode robust cluster. Metode Robust Clustering merupakan metode pengelompokan yang efektif dalam menghadapi outlier, memungkinkan identifikasi kelompok data yang homogen tanpa dipengaruhi oleh nilai ekstrem. Penelitian ini bertujuan untuk menggali dan mengatasi masalah data outlier dalam pemetaan dataset stunting. Berdasarkan hasil analisis clustering dengan metode K-Means, diperoleh tiga kelompok Puskesmas di Kabupaten Asahan berdasarkan jumlah kasus stunting tahun 2022–2024. Setelah dilakukan deteksi dan penanganan outlier, nilai Davies-Bouldin Index menurun dari 0,529 menjadi 0,483, menunjukkan kualitas cluster yang lebih baik. Hasil akhir menunjukkan bahwa mayoritas Puskesmas masuk ke dalam Cluster 1 dengan kasus stunting relatif rendah hingga sedang, sedangkan Cluster 2 terdiri dari Puskesmas dengan tren kasus meningkat cukup signifikan, dan Cluster 2 berisi Puskesmas dengan jumlah kasus sangat tinggi.
Â
Kata kunci: Stunting; Data Outlier; Robust Clustering; Pemetaan; Machine Learning
Downloads
References
De Onis M, Branca F. Childhood stunting: a global perspective. Maternal & Child Nutrition. 2016;12(S1):12-26.
Khan, L., & Hayat, A. (2024). Outlier detection in machine learning: A comprehensive review. Journal of Machine Learning Research, 25, 1-45.
Zhang, Z., & Zhao, X. (2024). Handling outliers in high-dimensional data: A new approach. Data Mining and Knowledge Discovery, 38(1), 12-34. doi:10.1007/s10618-023-00728-6.
Zhang C, Yang Q. A survey on multi-task learning. IEEE Trans Knowl Data Eng. 2023;35(1):1-20. doi:10.1109/TKDE.2022.3146592.
Gupta R, Sharma R. Enhancements in robust clustering methods for noisy data. Data Mining and Knowledge Discovery. 2024;38(1):45-67
Abeysinghe R, Abeywardena M, Ganewatta S. Applications of machine learning techniques in public health: A review. J Health Inform Dev Ctries. 2022;16(1):1-16.
Gonnelli S, Sassi F, Marini F. The Role of Machine Learning in Health Care: Applicability in Chronic Diseases. Int J Environ Res Public Health. 2021;18(3):1-15.
Alhussein M, Moderow T, Syed R. A comparative study of machine learning algorithms for assessing childhood malnutrition. J Nutr Health Food Eng. 2023;13(2):34-42.
Farhan M, Binsalman M, Beng H, Hussain M. Stunting prediction using advanced machine learning techniques: An analysis in rural settings. BMC Public Health. 2023;23:350.
Yulianto, Setiawati R. Leveraging tree-based models for the prediction of stunting among children: Evidence from Indonesia. Child Obes. 2023;19(1):10-16.
Lee J, Choi S. Exploring the intersection of machine learning and public health: A systematic review of current methods applied to stunting analysis. BMC Health Serv Res. 2023;23:558.
Aji P, Suharyanto A, Evita M. Random Forests for identifying socio-economic determinants of child stunting in Indonesia. Nutr Health. 2023;29(1):67-76.
Pramono M, Subekti A. The impact of economic factors on childhood stunting: An application of Gradient Boosting in rural Indonesia. J Dev Studies. 2023;59(4):650-670.
Farhan M, Binsalman M, Beng H, Hussain M. Stunting prediction using advanced machine learning techniques: An analysis in rural settings. BMC Public Health. 2023;23:350.
Abeysinghe R, Abeywardena M, Ganewatta S. Applications of machine learning techniques in public health: A review. J Health Inform Dev Ctries. 2022;16(1):1-16.
Yulianto, Setiawati R. Leveraging
