Estimating resource requirements of construction projects with artificial neural networks approach

ESTIMATING RESOURCE REQUIREMENTS OF CONSTRUCTIONPROJECTS WITH ARTIFICIAL NEURAL NETWORKS APPROACH(Ph.D. Thesis)Umut Naci BAYKANGAZİ UNIVERSITYINSTITUTE OF SCIENCE AND TECHNOLOGYJanuary 2007ABSTRACTIn this study, resource requirement, which is one the basic problems inconstruction sector, is focused on. For resource requirement neuralnetworks approach, one of the most interesting technologies of theartificial intelligence is used. Neural networks use samples to determinethe relationships between events, then gained relationships are used asa means of interpretation and decision making process for the solutionof problems that will be faced.In this context, project parameters are taught to neural networks andthe learning performance of the network is tested according to acquiredresults by illustrating the housing construction that have differentproject traits. Again these results are compared with the results whichare obtained by using same parameters in regression analyse.As a result, it is seen that, estimation performance of neural networks ishigher than regression analyse and also it is seen that estimations withhigh performance can be done by increasing the number of the sampleduring learning. It is stated that, this approach can be usedsuccessfully as a fast, productive and reliable method in resourceestimation of housing construction.Science Code : 714.3.035Key Words : Artificial neural networks, resource planning,resource estimatingPage Number : 117Adviser : Prof. Dr. Recep KANIT