Spatial Distribution of Poverty Clusters and Its Prediction Algorithms: A Visual Analytics Approach to Understanding the Disparities of Poverty Across Zones

Eradicating Poverty in all its forms remains a key priority and a significant challenge in Tanzania. More than 50% of the population lives in multidimensional poverty, with urban poverty stagnating while rural poverty, though higher, is declining at a slower rate. The challenge of higher multidimensional poverty, with a slow pace of reduction, entails ongoing empirical testing in order to inform strategies for achieving SDG 1 in 2030. This study aims to utilize a hybrid method to identify the most effective clustering algorithm for poverty distribution and its corresponding predictive algorithm at the household level across the different zones. The hybrid method was employed because supervised learning methods struggle with discovering hidden patterns due to pre-defined labels while clustering approach lacks predictive accuracy metrics. The analysis utilizes longitudinal data from the National Bureau of Statistics (NBS), where household surveys were conducted using a multi-stage, stratified cluster sampling technique. Multidimensional poor and non-poor households were classified using the K-Means clustering algorithm, which demonstrated the highest performance among the nine clustering algorithms tested, effectively categorizing households into two clusters. The findings indicate that while the overall poverty rate has declined slightly, it remains persistently high, affecting more than 65% of households in three zones: Western, Central and Lake zones. In contrast, the Coastal and Zanzibar zones show indications of an increasing number of poor households. The results reveal that the Ensemble Stacking method outperforms other predictive algorithms in accurately classifying poverty groups derived from K-Means clustering. In light of these findings, the study recommends that the government implement targeted policies that align with the specific poverty levels and socioeconomic conditions of each zone to enhance poverty alleviation efforts effectively.