İnşaat projelerindeki kalite uygunsuzluklarının maliyet etkilerinin CBR-AHP ve CBR-GA modelleriyle tahmin edilmesi

Quality problems in construction projects can have dramatic consequences, such as delays in timeline and cost overruns. If preventative measures are not applied, these quality problems can evolve into repetitive actions. This study introduces a proactive mechanism with the aim of protecting projects from the negative impacts of such non-conformities. It advances in three stages. The first involves determination of the attributes that comprehensively define quality problems using a literature survey and the Delphi method. The second determines the attribute weights by employing Analytical Hierarchy Process (AHP) and Genetic Algorithm (GA). In the final stage, a predictive model is developed to extract the possible outcomes of non-conformities in terms of cost impact. The predictive model adopts the case-based reasoning (CBR) approach with Mean Absolute Error (MAE) as the control criterion for prediction accuracy. Although CBR-GA yields a better MAE performance than CBR-AHP, the result is reversed for standard deviation. This thesis provides two significant outcomes in addition to the primary objective, forecasting possible failure. First, the attributes are determined to express the cases considered to contribute to the development of a record-keeping guideline for inexperienced quality practitioners. Second, the predictive model utilizes both automated and expert systems for attribute weighting, so the study examines the effect of automated and expert systems on the model's accuracy.