ANALISIS CLUSTERING INDEKS KEDALAMAN KEMISKINAN DI INDONESIA MENGGUNAKAN K-MEANS

Abstract

Poverty is one of the main issues in sustainable development that is measured using various indicators, one of which is the poverty depth index (P₁). This index provides a more comprehensive picture of the severity of poverty than the percentage of poor people (P₀). In this study, a clustering analysis of provinces in Indonesia is conducted based on time series data of the poverty depth index to identify similar patterns of change. This research applies the K-Means method with three distance measurement approaches, namely Dynamic Time Warping (DTW), Derivative Dynamic Time Warping (DDTW), and Weighted Dynamic Time Warping (WDTW). DTW is used to handle time shifts in the data, DDTW considers the pattern of change by calculating the first derivative, while WDTW assigns weights to each data point to produce more stable results. The clustering results were validated using the silhouette coefficient to determine the most optimal method. The results showed that the K-Means method with WDTW distance measurement had a silhouette coefficient value of 0.884 (strong).