The Analysıs Of Occupatınal Accıdents In Turkey Vıa Graph Theory In Data Mınıng Clusterıng Model

Thesis Type: Doctorate

Institution Of The Thesis: Gazi University, Turkey

Approval Date: 2017

Thesis Language: Turkish

Student: Gökhan Tuna

Supervisor: MUSTAFA KURT


Occupational accidents that directly affect the health and lives of employees are complex events, and this complexity has led to efforts to answer the question of how accidents happen. Mathematical visualization methods and numerical results are very useful in terms of research in social systems. In our thesis study, different from the quantitative methods frequently used in the analysis of the social network of occupational accidents, a subdivision of applied mathematics, graph theory is used. The aim of our work is, apart from the results of the methods used so far in the analysis of occupational accidents, to analyse the work accidents happened in Turkey in 2013-2014 using the graph theory for the discovery of unforeseen and unexpected new structures. In our work, each occupational accident was taken as a vertex of the graph and a graph structure was formed according to the correlation between a certain threshold value and each occupational accident after numerically specifying each parameter related to occupational accidents as a vertex coordinate. The pearson correlation coefficient was used to determine the relationship between the actors. The high modularity method used in clustering-community determination methods in the occupational accident social network structure. The topology of the sub-dominant ultra-metric structures in the occupational accidents was determined using the minimum spanning trees (MST) of the resulting graph and subgraphs and the hierarchical relationship between the events was analysed. The central vertices of the MST in the hierarchies were discovered as the most effective occupational accidents in the community corresponding to the local cluster. The independence tests were performed between the vectors of the central vertices in the lower communities, and the vertices binaries of highest dependence were found in the MSTs. Since the topology of the MSTs was established with the edges between actors, also topological measurement for the edges was presented. In such a high-dimensional analysis, graph theory provided a sense of the behaviour of the social network of occupational accidents that could be studied as a complex system.