Author: Zdravko Pečar
Menthor: prof. dr. Vladislav rajkovič
Comenthor: akademik prof. dr. Ivan Bratko
The dissertation Model for evaluating performance of work processes in public administration by using the methods of artificial inteligence presents interdisciplinary research by using machine learning programs and expert multivariable decision making program for assessing the quality. The sample of research are departments for spatial planning in 58 government administrative units, during the period 1996 – 1999. The research is mainly aimed at measuring the productivity in performing administrative services (planning permit, building permit, registering construction works, permit for use,...) and discovering how different attributes (size of administrative unit, amount of applications, number of employees, level of education) affect the change in efficiency and quality of administrative units and their services. For research data analyses, the sophisticated tools from the field of artificial inteligence were used to provide fast processing of large amounts of different data. The research resulted from several important findings: productivity of administrative units differs in ratio up to 1 : 10; the change in productivity is influenced by trends of new applications (increased number of applications increases productivity, and decreased number decreases it); higher level of education generaly yields higher productivity (the above average level of education often shows signs of lowering the productivity) and also gender and age of employees influence the productivity (the optimum is 6 – 24% men and the average age 39 – 42 years in employee structure).
Machine learning efficiently helps to discover different hidden patterns, among which some can benefit managerial decision making. By using the traditional methods this pathrns would remain unknown. The research also includes the application of another artificial inteligence tool for qualitative multi – attribute decision making DEXi in assessing the performance /quality of various administrative units based on developed model. The conclusions suggest the implementation of artificial inteligence methods in processes of decision making in public administration. The condition for successful implementation is improvement of existing performance measuring systems in public administration, as a source for adequate data bases.This developmental steps would pave the way toward performance based budgeting of public organisations and performance based pay of their employees.
Public administration, public management, performance measurement, performance management, statistical process control, artificial inteligence, knowledge discovery in data bases, machine learning, decision support systems.