Multivariate random forest prediction of poverty and malnutrition prevalence

Advances in remote sensing and machine learning enable increasingly accurate, inexpensive, and timely estimation of poverty and malnutrition indicators to guide development and humanitarian agencies’ programming. However, state of the art models often rely on proprietary data and/or deep or transfer learning methods whose underlying mechanics may be challenging to interpret. We demonstrate how interpretable random forest models can produce estimates of a set of (potentially correlated) malnutrition and poverty prevalence measures using free, open access, regularly updated, georeferenced data. We demonstrate two use cases: contemporaneous prediction, which might be used for poverty mapping, geographic targeting, or monitoring and evaluation tasks, and a sequential nowcasting task that can inform early warning systems. Applied to data from 11 low and lower-middle income countries, we find predictive accuracy broadly comparable for both tasks to prior studies that use proprietary data and/or deep or transfer learning methods.

Poverty Mapping in the Dian-Gui-Qian Contiguous Extremely Poor Area of Southwest China Based on Multi-Source Geospatial Data

Accurate information on the spatial distribution of poverty is of great significance to the formulation and implementation of the government’s targeted poverty alleviation policy. Traditional poverty mapping is mainly based on household survey data and statistical data, which cannot describe the spatial distribution of poverty well. This paper presents a study of mapping the integrated poverty index (IPI) in the Dian-Gui-Qian contiguous extremely poor area of southwest China. Based on multiple independent spatial variables extracted from NPP/VIIRS nighttime light (NTL) remote sensing data, digital elevation model (DEM), land cover information, open street map, and city accessibility data, eight algorithms were employed and compared to determine the optimal model for IPI estimation. Among these machine learning algorithms, traditional multiple linear regression had the lowest accuracy compared with the other seven machine learning algorithms and XGBoost showed the best performance. Feature selection was performed to reduce overfitting and five variables were finally selected. The final developed XGBoost model achieved an MAE of 0.0454 and an R2 of 0.68. The IPI map derived from the developed XGBoost model characterized the spatial pattern of poverty in the Dian-Gui-Qian contiguous extremely poor area well, which provided a good reference for the poverty alleviation work and public resources allocation in the study area. This study can also serve as a template for poverty mapping in other areas using remote sensing data.

Evaluating the Approaches of Small Area Estimation Using Poverty Mapping Data

Nowadays, estimation demand in statistics is increased worldwide to seek out an estimate, or approximation, which may be a value which will be used for various purpose, albeit the input data could also be incomplete, uncertain, or unstable. The development of different estimation methods is trying to provide most accurate estimate and estimation theory deals with finding estimates with good properties. The demand of small area estimation (SAE) method has been increasing rapidly around the world because of its reliability compared to the traditional direct estimation methods, especially in the case of small sample size. This paper mainly focuses on the comparison of several indirect small area estimation methods (poststratified synthetic, SSD and EB estimates) with traditional direct estimator based on a renowned data set. Direct estimator is approximately unbiased but SSD and Post-stratified synthetic estimator is extreme biased. To cope up the problem, we conduct another model-based estimation procedure namely Empirical Bayes (EB) estimator, which is unbiased and compare them using their coefficient of variation (CV). To check the model assumption, we used Q-Q plot as well as a Histogram to confirm the normality, bivariate correlation, Akaike information criterion (AIC). JEL classification numbers: C13, C51, C51. Keywords: Small Area Estimation, Direct Estimation, Indirect Estimation, Empirical Bayes Estimator, Poverty Mapping.

Poverty Reduction in Regencies/Municipalities in South Sumatra Province

Local Government expenditure is budgeting for all government needs and activities and managed under the authority of provinces, regencies, and municipalities through their respective regional heads. Well-targeted Local Government expenditure optimization has a significant impact on the regional economy. This research aims to determine poverty reduction in regencies/municipalities in South Sumatra Province, Indonesia, by examining the variable’s impact of social assistance expenditure, capital expenditure, and local revenue on poverty. The data used are primary and secondary data obtained from 15 regencies/municipalities in South Sumatra Province during the 2010-2018 periods. The analysis technique uses in this research were Poverty Mapping with Klassen Typology and Multiple Linear Regression (MLR). Using the Klassen typology for poverty mapping in South Sumatra Province obtained four regional classifications (quadrant) based on poverty and economic growth: quadrant I (developed and fast-growing region), quadrant II (developed but depressed region), quadrant III (developing region), and quadrant IV (less developed region). The Klassen typology classification results: quadrant I include Palembang City, quadrant II includes Musi Banyuasin Regency, Muara Enim Regency, Ogan Komering Ilir Regency, and Banyuasin Regency. Quadrant III includes Ogan Komering Ulu Regency, Prabumulih City, and Lubuk Linggau City. Also, quadrant IV includes Lahat Regency, Musi Rawas Regency, Ogan Ilir Regency, Ogan Komering Ulu Timur Regency, Ogan Komering Ulu Selatan Regency, Empat Lawang Regency, and Pagar Alam City. The t-test regression results showed that Social assistance expenditure and local revenue affect poverty reduction, while capital expenditure does not significantly affect poverty reduction. The F-test regression results showed that poverty reduction was affected simultaneously by social assistance expenditure, capital expenditure, and local revenue. Policies in social assistance expenditure and capital expenditure were not well-targeted. The policies expected to reduce poverty are to provide well-targeted social assistance expenditure and capital expenditure.

Poverty Mapping in Central Java Province Using K-Means Algorithm

Prosperity has a relative, dynamic, and quantitative meaning. Until now, the formula is not finished because it will continue to grow along with the times. Public welfare is a condition where all citizens are always in a condition that is completely adequate in all their needs. Poverty in Central Java Province is still above national poverty. Poverty grouping is one way to focus on the people's budget in each region so that they can take development policies and strategies that are right on target and effective. In this study, the proposed K-means algorithm for classifying poverty in Central Java is based on poverty indicators. The results of the first cluster study consisted of 22 districts / cities with the category of not poor, the second cluster consisted of 13 districts / cities that were categorized as poor.

MODEL PENGENDALIAN KEMISKINAN DENGAN PENDEKATAN DIVERSIFIKASI USAHA, SUSTAINABLE DEVELOPMENT GOALDS (SDGS) DAN ECONOMIC VALUE (STUDI PADA DAERAH PEMEKARAN DI SUMATERA UTARA)

This research has the goal of providing input to economic divesifikasi models which are the most consistent in increasing economic value and decrease poverty in the district town expansion North Sumatra Province. Provide input model Sustaianble Development Goals (SDGs) are the most supportive in the increase in economic value and alleviating poverty in the county town expansion North Sumatra Province. Prove empirically the performance of the economy which interact strongly supported the decline in poverty in the county townof the expansion in North Sumatra Province. This research uses 16 sample extraction area in North Sumatra. The first model of poverty alleviation through diversified value-added business community in meeting the needs of the community's life while the second model to determine what variables as policy recommendations for alleviating poverty in the County the expansion North Sumatra Province. The first model requires primary data while the second model requires data skunder time series starting in 2000 up to 2013. data analysis using two models, namely Structural Equation Models (SEM) and Vector Autoregression (VAR). Results of the study and this analysis gives an overview and prove the real poverty mapping in areas of expansion, poverty reduction actions based on the poverty mapping, control variables are very urgent in affects poverty, and controlling poverty continuously.

Multi-View Learning from Disparate Sources for Poverty Mapping

Many data analytics problems involve data coming from multiple sources, sensors, modalities or feature spaces, that describe the object of interest in a unique way, and typically exhibit heterogeneous properties. The varied data sources are termed as views, and the task of learning from such multi-view data is known as multi-view learning. In my thesis, I target the problem of poverty prediction and mapping from multi-source data. Currently, poverty is estimated through intensive household surveys, which is costly and time consuming. The need is to timely and accurately predict poverty and map it to spatially fine-grained baseline data. The primary aim of my thesis is to develop novel multi-view algorithms that combine disparate data sources for poverty mapping. Another aim of my work is to relax the core assumptions faced by existing multi-view learning algorithms, and produce factorized subspaces.