Classification and regression tree theory application for assessment of building damage caused by surface deformation

Abstract

A framework of applying the classification and regression tree theory (CART) for assessing the concrete building damage, caused by surface deformation, is proposed. The prognosis methods used for approximated building hazard estimation caused by continuous deformation are unsatisfactory. Variable local soil condition, changing intensity of the continuous deformation and variable resistance of the concrete buildings require the prognosis method adapted to the local condition. Terrains intensely induced by surface deformation are build-up with hundreds of building, so the method of their hazard estimation needs to be approximated and relatively fast. Therefore, promising might be addressing problems of reliable building damage risk assessment by application of classification and regression tree. The presented method based on the classification and regression tree theory enables to establish the most significant risk factors causing the building damage. Chosen risk factors underlie foundation for the concrete building damage prognosis method, which was caused by the surface continuous deformation. The established method enabled to assess the severity of building damage and was adapted to the local condition. High accuracy of shown approach is validated based on the independent data set of the buildings from the similar region. The research presented introduces the CART to determination of the risk of building damage with the emphasis on the grade of the building damage. Since presented method bases on the observations of the damages from the previous subsidence, the method might be applied to any local condition, where the previous subsidence is known.

A comparative study of logistic model tree, random forest, and classification and regression tree models for spatial prediction of landslide susceptibility

“…Classification and regression tree is a recursive partitioning method, which builds classification and regression trees for predicting categorical predictor variables (classification) and continuous dependent variables (regression) (Felicísimo et al, 2013;Youssef et al, 2015b). This method is widely used in many fields (Bevilacqua et al, 2003;Kim et al, 2015;Koon and Petscher, 2015;Malinowska, 2014;Yang et al, 2016). The CART is constructed by splitting subsets of the dataset using all predictor variables to create two child nodes repeatedly, and the final goal is to produce subsets of the dataset which are as homogeneous as possible with respect to the target variable (Mahjoobi and Etemad-Shahidi, 2008).…”

Effects on Buildings of Surface Curvature Caused by Underground Coal Mining

“…The research introduced CART to determine the risk of building damage, with the emphasis on the grade of building damage. Given that the presented method is based on observations of damage from the previous subsidence, the method can be applied to any local conditions in which the previous subsidence is known [7]. Malinowska employed a fuzzy inference method coupled with GIS to enable the integration of diverse factors that affect risk such as surface deformations and resistance of building objects, with consideration of data uncertainty and subjectivity of evaluation of experts that make the assessment.…”

K-Nearest Neighbours Method as a Tool for Failure Rate Prediction

“…support vector machine -SVM, regression tress -RT and K-nearest neighbours -KNN) were used to solve many engineering problems. For instance, the localization of leakages from water-pipe network was estimated using SVM algorithm [14], the building damage was assessed by means of RT [15] and KNN algorithm was used relating to time series analysis in industrial processes [16]. The main aim of this paper is to check if nonparametric regression algorithm KNN could be also useful for prediction of indicator λ of water conduits (water mains, distribution pipes and house connections).…”