| Title |
A Study on the Correlation between Selection Methods of Input Variables and Number of Data in Estimation Accuracy |
| Authors |
Kim Gwang-Hee ; An Sung-Hoon |
| Keywords |
Apartment Housing Project ; Cost ; Neural Networks ; Genetic Algorithm ; Input Variable ; Number of Data |
| Abstract |
The purpose of this study is to examine the effects of the selection methods of input variables and the number of data on the accuracy of cost estimation using neural networks(NNs) in apartment housing projects. The acquired 10 input variables was classified into three categories(concept, pre-design, preliminary design stage). And then input variables for constructing the neural networks model was selected by regression analysis, principal component analysis(PCA), genetic algorithms. The neural network models were constructed by changing the number of data and the number of input variables according to the method for selecting optimal input variables and project definition. Using the NNs for cost estimation, the method that selects the input variables by regression analysis could enhance the accuracy. The relationship also could reduce time because the range of accuracy is expected |