Development of a group decision making method for ranking alternatives: Selection of most preferred data mining algorithm for a construction project

This thesis develops a methodology for selecting the most preferred data mining algorithm for a construction project. The proposed methodology exploits the Analytical Hierarchy Process (AHP). AHP has been widely applied to a variety of complex problems. The contributions of this thesis are twofold. The first part is the design and development of the methodology. To use AHP for data mining, several modifications must be made. The most important of which is to secure a collective decision-making environment for a group of different backgrounds and an appropriate data collection method for AHP The second part of the contribution is the application to the problem of selecting the most preferred data mining algorithm for a construction project. Where it addressed the limitations that were identified in the literature, by taking all performance measures into account, considering the knowledge of field experts about the relative importance of these measures and saving time and resources with the possibility of making a decision when there is no data from testing the models. The structure of the hierarchical model is designed so that each participant can make judgments only in their area of specialization without having to understand the model as a whole. This feature allows adding different experiences that cover the different dimensions of the problem. The data for this study was collected from two groups of participants, one group consisted of construction practitioners, and the second group consisted of machine learning practitioners. The data collected are the personal preferences of each participant.