A decision support software developed with machine learning in selection of a demolition method according to structural and environmental features of building

Building demoliton is considered as a multi-criteria decision problem. Past experiences of demolition engineers come to the fore in solving this problem based on the characteristics of the structures and their surroundings. The selection of technique should be decided by considering the criteria in the demolition regulations applied in our country as well as the past experiences of the experts. Within the scope of this thesis, machine learning algorithms have been used in the selection of demolition techniques because it can learn from previous data. The data set to be used for this purpose was produced synthetically since it could not be obtained from previous studies. In the studies of producing the data set, the factors in the demolition regulations applied in our country and the data verified in line with the opinions of demolition experts were used. As a result of the machine learning training processes carried out with these data, a model was created, integrated with the developed web-based software, and a demolition technique was proposed to experts according to the characteristics of the buildings to be demolished. With the software, building properties data and suitable demolition technique recommendations are received from experts. A system with increasing learning rate is targeted by operating machine learning processes with data from experts. It is thought that with the development of the data set, it will guide researchers who want to work in this field in our country.