Modelling Investment Programs With Machine Learning and Data Mining: Descriptive and Predictive Models of Healthcare State Programs in Russia

Proceedings of the XXth Conference of Open Innovations Association FRUCT

卷:31

Georgy Kopanitsa¹  Oleg Metsker¹  Aleksandr Gerasimov²  Egor Trofimov³ 

[1] ITMO University, Russia;

[2] National Research UniversityHigher School of Economics, Russia;

[3] The All-Russian State University of Justice, Russia;

关键词: big data;    computer analytics;    governance;    e-governance;    policy modelling;   

DOI  :  10.23919/FRUCT54823.2022.9770894

来源: DOAJ

【 摘 要 】

This paper describes the results of data research and the development of models based on machine learning for forecasting and describing factors affecting the achievement of federal projects on the example of a project to improve the quality of life in Russia. We searched for the best predictors to forecast an infant mortality as one of the direct indicators of a healthcare system quality. The dataset included the social and economical indexes of Russia. We perfumed a Gradient Boost with 0.98 R_squared. We performed a correlation analysis using F-Measure to find the predictors with the highest predictive power. During the research work, it was possible to identify factors affecting the indicators of the national health project. It is worth noting that the indicators of national projects are influenced by the level of education of the head of the region and work experience. The combination of a large territory, a high proportion of the urban population and economic activity are three important factors that require the modernization of the system of public administration. How-ever, in conditions, when public administration is highly unified, such a transformation practically does not take place. This work certainly contributes to the development of methods of applying machine learning to analyze complex performance indicators of public administration. The conclusion is made about the possibilities of using the obtained model to supplement the program-objective methods used in government programs with more fine and precise regulatory tools to achieve greater efficiency by optimizing the law.

【 授权许可】

Unknown