Applying machine learning methods for analysis socio-economic survey data

Abstract

To ensure the content of decision-making information systems in the individuals’ assets management process requires the development of mathematical models of complex social systems. Studying the expectations of young people on socio-economic issues is of great importance for understanding the future development of the state for developing social policy strategies. A priority throughout the life cycle of an individual is happy and stable marriage, for the stability of which material and moral well-being is important. The next important factor of growing up is closely related to solving the housing problem. The strategic goal of most universities is to train highly paid specialists who are capable to develop the country and support the well-being of own family. Planning expected income is one of the steps of the family welfare planning algorithm. The purpose of this work is to study factors that are important for maturation and well-being. Using machine learning methods, the work explores socio-economic problems from the point of view of first-year university students. The influence of various factors for making decisions regarding the expected age of marriage, solving the housing problem, and expected job income is considered in the research work. Pre-processing of survey data applies data mining techniques. A comparative analysis of the forecast accuracy of classification methods is carried out: logistic regression, neural networks, support vector machines. Students are clustered using the K-means method.