<![CDATA[Poverty is still one of the main problems in economic development and inequality, unemployment, and economic growth. This study aims to model poverty directly by using a discrete choice model using binomial regression. The data used is imbalanced data, where one of the value categories is relatively small. In this study, the logistic regression method applies several resample techniques. They include undersampling, oversampling, a combination of both, and Cost-Sensitive Learning (CSL). The results obtained that both sampling techniques provide optimal results when viewed from the indicators of accuracy, specificity, sensitivity, and AUC. In addition, the results show that in households in rural areas, the head of the household is female, unmarried, has low education, married at an early/old age, and has a large household size, has a greater chance of being poor than other categories. So that targeted and comprehensive policy is needed so that the poverty rate can continue to be reduced and welfare increases]]>
Copyrights © 2021
