Exploring Socio-Economic Determinants of Poverty in Somalia: Insights from Descriptive Statistics and Machine Learning Models

The analyzed dataset from the Socioeconomic and Demographic Health Survey (SDHS) 2020 in Somalia provides valuable insights into the determinants of poverty in the country. By examining various categorical and continuous variables, researchers were able to identify key factors associated with poverty and develop predictive models to aid poverty alleviation efforts.

The study revealed that factors such as educational attainment, gender dynamics, access to basic amenities, and geographical location play significant roles in determining poverty levels in Somalia. For instance, individuals with higher levels of education had higher odds of poverty, while female-headed households exhibited lower odds of poverty. Additionally, households with unimproved water sources and toilet facilities were more likely to experience poverty.

Machine learning models, including Random Forest (RF), Decision Trees (DT), Support Vector Machines (SVM), and logistic regression, were employed to predict poverty levels in the population. The RF model outperformed the other models in terms of accuracy, recall, sensitivity, specificity, and other performance metrics. The model’s ability to accurately classify individuals as poor or well-off was reflected in its high area under the receiver operating characteristic curve (AUROC) value.

Furthermore, the feature selection process using the RF model identified 13 key features associated with poverty, including administrative region, household size, age group of respondents, employment status of the husband, and access to basic amenities. These findings highlight the multifaceted nature of poverty in Somalia and emphasize the importance of considering various socio-economic factors in poverty alleviation strategies.

Overall, the study’s comprehensive analysis of descriptive statistics, inferential analysis, machine learning models, and feature selection provides valuable insights into poverty dynamics in Somalia. By leveraging data-driven approaches and predictive modeling, researchers aim to inform targeted interventions and policies to address poverty and improve socio-economic conditions in the country.