로지스틱 회귀분석을 이용한 적정 지하 터파기 공법선정 의사결정 지원 시스템

The increased importance of excavation work in urban construction projects leads many researchers to try to develop prediction models using various data mining techniques that can help construction engineers select appropriate retaining wall systems. However, the proposed prediction models showed flaws most likely attributed to the insufficient number of data. Since collecting sufficiently a large number of data is nearly impossible, we propose an alternate approach that uses logistic regression to overcome the problems of a limited number of data. 139 excavation cases were analyzed by a series of logistic regression, quantifying statistically significant relations between site conditions and respective retaining wall systems. These relations were reviewed again in practical perspective. Both statistically significant and practically important relations were then developed into the decision support process, a decision tree-like structure. The decision-support process proposed in this study showed around an 80% accuracy rate, which is higher than that of the decision tree built by an automated machine learning algorithm.