Machine Learning for Policy Guidance

Abstract

This paper leverages machine learning algorithms and techniques to create models that can assist in a country's policy guidance. The machine learning process used to conduct research is discussed with steps such as preprocessing, feature selection, model selection, and model interpretation. Specifically, using datasets from the CIA's World Factbook and the United Nations' Human Development Index (HDI), machine learning models are created that use select features from several counties (e.g., real gross domestic product (GDP), population, and area). Then, the models make predictions on the countries' HDI scores. Model interpretation methods are used to find the most important features in predicting a country's score. This paper argues that important features can be derived through machine learning and guide government policy relevant to human development. Supply-side policies are discussed based on the results from the machine learning models. The use of machine learning with other indexes is also explored.