Prediction of occupational accidents, risk analysis and prioritization of precautions to be taken by house of quality in construction industry

Occupational accidents are complex and serious problems that cannot be prevented at the desired level. The issues of data-driven analysis and prediction of occupational accidents, risk analysis and prioritization of precautions to be taken are considered within the scope of this thesis study. These issues are examined in the construction industry, where occupational accidents occur frequently in our country. Occupational accidents data occurred in construction workplaces in the Central Anatolia Region between 2013-2017 are obtained from the Social Security Institution (SSI) for the analysis and prediction of occupational accidents, statistical analyzes of the data are carried out and prediction models are developed for the working days lost and accident outcome as accident severities. Binary Logistic Regression with Firth, Ridge, Lasso and Elastic net regression models are generated for the prediction of working days lost and Multinomial Logistic Regression with Ridge, Group Lasso and Group Elastic net regression models are generated for the predicton of accident outcome. A risk analysis study is conducted for a multi-storey building construction project with an Interval-Valued Neutrosophic AHP and Interval-Valued Neutrosophic CoCoSo integrated Multi-Criteria Decison Making (MCDM) model, which is developed based on the Fine-Kinney method. K-means clustering approach is utilized to determine the risk classes of hazards. The prioritization of precautions is provided by associating the hazards determined to be in the very high risk category with the precautions to be taken in the Interval-Valued Neutrosophic House of Quality model. In this study, the effectiveness of all the models developed for the prediction of occupational accidents, risk analysis and prioritization of precautions to be taken are tested and good results are revealed.