Forecasting Indonesia's poor population province using machine learning algorithm analysis.

Poverty is still one of the main problems for developing countries including Indonesia. Poverty is a multidimensional problem not only in terms of income but also in other factors such as health, education, access to goods and services, location, geographical conditions, gender, and environmental conditions. Poverty is not only the responsibility of the central government but also the respective regional governments and all levels of society. Therefore, this study was conducted as a contribution to the government in providing information about the forecasted number of poor people in 2023. This study used a dataset of the number of poor people in Indonesia by province from 2015 to 2022, sourced from the National Socioeconomic Survey. The proposed algorithm was Machine Learning with the Quasi-Newton BFGS technique using Matlab version 2011b. The dataset was analyzed using eleven architectural models: 6-3-1, 6-4-1, 6-5-1, 6-8-1, 6-10-1, 6-11-1, 6-15-1, 6 -20-1, 6-25-1, 6-5-5-1, and 6-11-1. After being analyzed, the best model obtained was 6-5-1 due to the lowest MSE and RMSE values compared to the other models. Therefore, the 6-5-1 architectural model can be used to forecast the number of poor people in Indonesia by the province in 2023. Thus, based on the forecasting results, the poor population in Indonesia in 2023 is expected to decrease compared to 2022. [ABSTRACT FROM AUTHOR]