| Title |
Prediction Performance Analysis of Total Construction Period through Normalization of Independent and Dependent Variables |
| Authors |
강윤호(Kang, Yun-Ho) ; 이하늘(Lee, Ha-Neaul) ; 윤석헌(Yun, Seok-Heon) |
| DOI |
https://doi.org/10.5659/JAIK.2023.39.11.281 |
| Keywords |
Construction Period; Log; Machine Learning; Normalization |
| Abstract |
The estimation of the appropriate construction period in the construction project is one of the main factors for successfully completing the
project. Various artificial intelligence technologies have been developed and various efforts have been made to improve prediction performance
during the construction period by applying them, but it is still difficult to accurately predict the construction period. It is judged that the
predictive performance of the learning data during the construction period may be poor due to the large deviation of the data between the
independent variable and the dependent variable. In this study, it was intended to improve the predictive performance of the construction
period by reducing the data deviation through the normalization of the dependent and independent variables of the training data. In this
study, a total of 953 data from by the PPS(Public Procurement Service) in Korea were used for five years from 2017 to 2022, and to
reduce the relative difference between independent and dependent variable data, three models were defined, and the training results were
compared and analyzed. As a result of the analysis, it is judged that the model using Log normalization has the best prediction performance. |