Analysis of Multidimensional Poverty Indicators in Indonesia with Association Rules

Abstract

This study was conducted to find patterns of relationships between 14 multidimensional poverty indicators in Indonesia from 2015-2019. To provide a more specific description of the relationship pattern, association rules with the apriori algorithm is used as the analysis method. The preprocessing stage to transform data was carried out using fuzzy functions and data reduction with Multiple Correspondence Analysis (MCA) to support the association analysis process. The results obtained are 15 relationship patterns or rules between items from the multidimensional poverty indicator with a support value of 60%-80% and 100% confidence. This means that the relationship pattern is significantly formed from objects with a strong relationship between the items and can represent poverty records in the last five years. The relationship pattern consists of four combinations of things. Suppose there is a high category decrease in the percentage of poor people indicator, a low category decrease in the open unemployment indicator, a high category increase in the percentage of households indicator according to the source of lighting from electricity, and a low category increase in the percentage indicator of households according to the broadest wall, not bamboo / other. In that case, there is a reduction in multidimensional poverty in Indonesia.