engleski

Hybrid approach based on multi-criteria decision making and machine learning in the public sector

In the process of making decisions in the public sector, various methods can be used, but practitioners face many problems in the implementation: (1) the methods are very complex, (2) the process is time-consuming, and (3) there is a lack of the experts that should be involved, and (4) users are limited in resources. In this paper, we are proposing a new, hybrid approach to make decisions in the public sector. The approach combines methods from the areas of multi-criteria decision making (SNAP and composite index) and machine learning (artificial neural network. Some of many benefits of the new approach are: (1) once created, the decision model can be reusable without need that users know the details of the methodology, (2) the proposed model considers interactions between elements of the model, without issues in stochasticity of problem matrix (like in ANP), (3) since the neural network model contains experts’ knowledge, it is not needed that in future implementations of the model we repeatedly include new (or same) experts again ; and (4) the model contains long-term knowledge about the criteria and alternatives of the problem and that is often a request in public sector problems.

Public sector ; SNAP ; composite index ; artificial neural network ; efficiency

nije evidentirano