| Title |
A Study on Stochastic Gradient Tree Boosting Model for the Retaining Wall Selection in Domestic Building Construction |
| DOI |
https://doi.org/10.5659/JAIK.2021.37.7.195 |
| Keywords |
Stochastic gradient tree boosting (SGTB); excavation work; retaining wall; building construction |
| Abstract |
The rise of land prices and population density in urban areas has led to a need for deeper excavations, both for the building ground and the
underground. It is difficult to select a retaining wall method that is appropriate for a construction site, not only because the retaining wall
method should be chosen at an early stage of a construction project, at which time there is a lack of information on surrounding
characteristics of the site, but also because there are uncertain factors such as underground water and the underlying rock formation. An
inappropriate retaining wall method may cause changes in the original design or method of retaining wall, resulting in an inevitable increase
in construction costs. Despite this fact, construction practitioners generally select a retaining wall method depending on their own limited,
subjective experience and intuition. For this reason, in this study, I applied the stochastic gradient tree boosting (SGTB) technique to
selecting a retaining wall method to assess the applicability of the technique to a work method selection. To evaluate the SGTB technique's
performance, I built the models using NN as well as SGTB and then compared the results between the models. As a result, it was found
out that the SGTB is relatively more excellent and stable compared to NN model when it comes to selecting a retaining wall. Consequently
SGTB is helpful to practitioners who need to determine the excavation work at building construction project. |