A Comparison of Machine Learning Techniques for Analyzing Infant Mortality in Bangladesh

The Infant Mortality Rate (IMR) is a significant measure used to evaluate the health and socioeconomic status of developing nations, particularly Bangladesh. It is important to carefully examine the various risk factors associated with IMR. This study aims to identify potential risk variables that contribute to infant mortality in Bangladesh and utilize machine learning techniques to develop prediction models. We have analyzed data from the World Bank's “World Development Indicator (WDI) 2021” repository. By employing multiple stepwise linear regression, Support Vector Machine, and Random Forest algorithms, we have identified relevant components and modeled the data. Through cross-validation techniques and evaluating performance using Mean Absolute Error, Mean Square Error, and Root Mean Square Error, we have determined that linear regression and support vector regression show superior performance. The final prediction model's effectiveness yields significant results, potentially leading to a significant reduction in Bangladesh's infant mortality rate significantly as well as improving the country's socio-economic conditions among other developing nations worldwide.