Widyawati Widyawati, Muhamad Oki Astrabuwono

Pengelompokan Kabupaten/Kota di Provinsi Banten Menggunakan K-Means Clustering Berbasis Indikator Pembangunan Multidimensi untuk Penentuan Prioritas Program Daerah

Introduction

Pengelompokan kabupaten/kota di provinsi banten menggunakan k-means clustering berbasis indikator pembangunan multidimensi untuk penentuan prioritas program daerah. Pengelompokan kabupaten/kota Banten dengan K-Means clustering berdasarkan indikator pembangunan multidimensi. Identifikasi 3 cluster: prioritas intervensi, berkembang, maju. Rekomendasi kebijakan program daerah terarah.

Abstract

Penelitian ini bertujuan untuk mengelompokkan kabupaten/kota di Provinsi Banten berdasarkan indikator pembangunan multidimensi, menganalisis karakteristik setiap cluster, serta memberikan rekomendasi awal mengenai prioritas program daerah. Penelitian ini menggunakan pendekatan kuantitatif deskriptif dengan data sekunder yang bersumber dari publikasi resmi Badan Pusat Statistik Provinsi Banten tahun 2024. Objek penelitian mencakup 8 kabupaten/kota di Provinsi Banten dengan enam variabel, yaitu garis kemiskinan, tingkat pengangguran terbuka, pertumbuhan ekonomi, produk domestik regional bruto atas dasar harga konstan (PDRB ADHK), rata-rata lama sekolah, dan umur harapan hidup. Data dianalisis melalui tahapan normalisasi Min-Max dan implementasi algoritma K-Means Clustering menggunakan Python. Hasil penelitian menunjukkan bahwa kabupaten/kota di Provinsi Banten terbagi ke dalam tiga cluster, yaitu wilayah prioritas intervensi, wilayah berkembang, dan wilayah maju. Hasil evaluasi menunjukkan nilai Silhouette Score sebesar 0.3216 yang mengindikasikan kualitas cluster yang cukup. Hasil pengelompokan ini menegaskan adanya perbedaan karakteristik pembangunan antarwilayah, sehingga kebijakan pembangunan daerah tidak dapat diterapkan secara seragam. Penelitian ini berkontribusi dalam menyediakan pemetaan pembangunan daerah berbasis data multidimensi yang dapat digunakan sebagai dasar awal untuk mendukung penyusunan kebijakan pembangunan yang lebih terarah, objektif, dan sesuai dengan kebutuhan masing-masing wilayah.

Review

The paper titled "Pengelompokan Kabupaten/Kota di Provinsi Banten Menggunakan K-Means Clustering Berbasis Indikator Pembangunan Multidimensi untuk Penentuan Prioritas Program Daerah" presents a relevant and timely study aimed at classifying districts/cities in Banten province based on multidimensional development indicators. The objective of providing an initial recommendation for regional program priorities is highly commendable, as it addresses the need for tailored development policies. Utilizing a quantitative descriptive approach with secondary data from BPS, the research employs K-Means clustering on six pertinent variables, including economic, social, and human development metrics. The identification of three distinct clusters—wilayah prioritas intervensi, wilayah berkembang, and wilayah maju—is a significant finding that underscores the heterogeneity of development across the region, thereby reinforcing the argument against uniform policy application. The use of Python for implementation and the calculation of a Silhouette Score demonstrate a systematic and data-driven methodological approach. While the study provides valuable insights, there are areas that could be further elaborated or strengthened. The Silhouette Score of 0.3216, though deemed "cukup," suggests that the clusters might not be extremely well-separated or distinct, which warrants a deeper discussion on the robustness of the clustering results. It would be beneficial to know how the optimal number of clusters (k=3) was determined; whether methods like the Elbow method, Gap statistic, or a clear justification based on domain knowledge were employed. Furthermore, while the abstract mentions analyzing the characteristics of each cluster, a brief preview of these distinct characteristics for each cluster (e.g., which variables heavily influence each cluster's identity) would enhance the abstract's descriptive power. Clarification on the data year "2024" (if it refers to data *published* in 2024 or data *collected* in 2024, which would be future data) is also needed for precision. Despite these minor points, the research offers a substantial contribution by providing a data-driven mapping of regional development, which is crucial for fostering more targeted and objective policy-making. The findings serve as a robust initial basis for local governments in Banten to design development programs that are genuinely responsive to the unique needs and challenges of each clustered region. For future research, it would be valuable to delve deeper into specific policy recommendations for each cluster, perhaps by integrating qualitative insights or stakeholder consultations. Exploring the temporal dynamics of these clusters over several years could also reveal important trends and the effectiveness of past interventions. Additionally, a comparative analysis using alternative clustering algorithms or a sensitivity analysis regarding variable selection could further validate and enrich the study's conclusions, strengthening its potential impact on regional planning and resource allocation.

Full Text

Comments

You need to be logged in to post a comment.