신경망과 유전자알고리즘을 이용한 공사비예측 모델의 예측정확도 비교에 관한 연구

The purpose of this study is to compare the accuracy of cost prediction models based on neural networks and genetic algorithm in apartment housing project at the project planning stage. Data used in this study are apartment houses' cost constructed from 1997 to 2000 in Seoul. The models applied were Model I, II, and III. The Model I is constructed by a trial-and error process , that is, the parameters of neural network were determined by a trial-and error process. The parameters of Model II is determined by genetic algorithm. And the Model III, which trained weights of neural networks using genetic algorithm. In this study, in case of determining the parameters of neural networks using genetic algorithm, the accuracy of cost prediction is higher than others in predicting apartment housing project at the project planning stage.