Neural network modeling of motivation for government agencies’ top management of regions as a classification problem

Abstract

Subject. The study deals with modeling the motivation of top managers of government agencies in regions to align the interests of people and the State. Objectives. The purpose of the study is to create a neural network model of motivation for top management of regional government institutions for a classification problem. Methods. Using neural networks, we simulate criteria for non-financial and financial motivation of the said top management, and criteria for strategic potential of regions. Financial motivation is defined as the salary of a senior civil servant, and non-financial motivation as his or her career growth. At the same time, the target function is a coefficient of natural population growth in regions, its positive value is assessed positively, and negative value negatively. As a result, the problem of binary classification in the trained neural network is solved. Results. Comparing the accuracy of the model in the considered example with accuracy that was obtained earlier, using logistic regression, we note that in the previous model, the total error in verification by the functions of non-financial and financial motivation and strategic potential was 39%. In our case, this error was only 12%. This suggests that neural networks enable to achieve much more accurate forecasting. Conclusions. The findings could be useful for regional government agencies to develop a constructive system of non-financial and financial motivation for their top managers.