Performance of Logistic Regression and Support Vector Machines for Seismic Vulnerability Assessment and Mapping: A Case Study of the 12 September 2016 ML5.8 Gyeongju Earthquake, South Korea
Abstract:In this study, we performed seismic vulnerability assessment and mapping of the ML5.8 Gyeongju Earthquake in Gyeongju, South Korea, as a case study. We applied logistic regression (LR) and four kernel models based on the support vector machine (SVM) learning method to derive suitable models for assessing seismic vulnerabilities; the results of each model were then mapped and evaluated. Dependent variables were quantified using buildings damaged in the 9.12 Gyeongju Earthquake, and independent variables were co… Show more
“…Thus, several studies have found that tree-based machine learning models exhibited higher performance than statistical models, and RF models exhibited high performance in most studies, confirming their suitability for vulnerability analysis. In a previous study, Han et al (2019) [20] used 15 factors except for social indicators to build LR and SVM kernel models to compare and analyze their performance. The results showed that the performance of the model based on the radial basis function (RBF) kernel (0.998) of SVM was the best, followed by polynomial (0.842), linear (0.649), LR (0.649), and sigmoid (0.630).…”
Section: Discussionmentioning
confidence: 99%
“…Many recent studies related to seismic vulnerability assessment and mapping have been conducted using machine learning techniques [12,[18][19][20][21]. For example, Han et al (2019) [20] used a logistic regression (LR) model and applied the support vector machine (SVM) methodology to four kernel models (linear, polynomial, radial basis function, and sigmoid) to derive a suitable model for seismic vulnerability assessment; this study was notable in that the results of several seismic vulnerability models were compared analytically; such analyses are rarely conducted in this field, despite the broad application of machine learning techniques in recent years.…”
The main purpose of this study was to compare the prediction accuracies of various seismic vulnerability assessment and mapping methods. We applied the frequency ratio (FR), decision tree (DT), and random forest (RF) methods to seismic data for Gyeongju, South Korea. A magnitude 5.8 earthquake occurred in Gyeongju on 12 September 2016. Buildings damaged during the earthquake were used as dependent variables, and 18 sub-indicators related to seismic vulnerability were used as independent variables. Seismic data were used to construct a model for each method, and the models’ results and prediction accuracies were validated using receiver operating characteristic (ROC) curves. The success rates of the FR, DT, and RF models were 0.661, 0.899, and 1.000, and their prediction rates were 0.655, 0.851, and 0.949, respectively. The importance of each indicator was determined, and the peak ground acceleration (PGA) and distance to epicenter were found to have the greatest impact on seismic vulnerability in the DT and RF models. The constructed models were applied to all buildings in Gyeongju to derive prediction values, which were then normalized to between 0 and 1, and then divided into five classes at equal intervals to create seismic vulnerability maps. An analysis of the class distribution of building damage in each of the 23 administrative districts showed that district 15 (Wolseong) was the most vulnerable area and districts 2 (Gangdong), 18 (Yangbuk), and 23 (Yangnam) were the safest areas.
“…Thus, several studies have found that tree-based machine learning models exhibited higher performance than statistical models, and RF models exhibited high performance in most studies, confirming their suitability for vulnerability analysis. In a previous study, Han et al (2019) [20] used 15 factors except for social indicators to build LR and SVM kernel models to compare and analyze their performance. The results showed that the performance of the model based on the radial basis function (RBF) kernel (0.998) of SVM was the best, followed by polynomial (0.842), linear (0.649), LR (0.649), and sigmoid (0.630).…”
Section: Discussionmentioning
confidence: 99%
“…Many recent studies related to seismic vulnerability assessment and mapping have been conducted using machine learning techniques [12,[18][19][20][21]. For example, Han et al (2019) [20] used a logistic regression (LR) model and applied the support vector machine (SVM) methodology to four kernel models (linear, polynomial, radial basis function, and sigmoid) to derive a suitable model for seismic vulnerability assessment; this study was notable in that the results of several seismic vulnerability models were compared analytically; such analyses are rarely conducted in this field, despite the broad application of machine learning techniques in recent years.…”
The main purpose of this study was to compare the prediction accuracies of various seismic vulnerability assessment and mapping methods. We applied the frequency ratio (FR), decision tree (DT), and random forest (RF) methods to seismic data for Gyeongju, South Korea. A magnitude 5.8 earthquake occurred in Gyeongju on 12 September 2016. Buildings damaged during the earthquake were used as dependent variables, and 18 sub-indicators related to seismic vulnerability were used as independent variables. Seismic data were used to construct a model for each method, and the models’ results and prediction accuracies were validated using receiver operating characteristic (ROC) curves. The success rates of the FR, DT, and RF models were 0.661, 0.899, and 1.000, and their prediction rates were 0.655, 0.851, and 0.949, respectively. The importance of each indicator was determined, and the peak ground acceleration (PGA) and distance to epicenter were found to have the greatest impact on seismic vulnerability in the DT and RF models. The constructed models were applied to all buildings in Gyeongju to derive prediction values, which were then normalized to between 0 and 1, and then divided into five classes at equal intervals to create seismic vulnerability maps. An analysis of the class distribution of building damage in each of the 23 administrative districts showed that district 15 (Wolseong) was the most vulnerable area and districts 2 (Gangdong), 18 (Yangbuk), and 23 (Yangnam) were the safest areas.
“…The FR model applied in this study is based on the assumption that future seismic activity in the study area will affect the area and the area's vulnerability to the future seismic event/s is directly correlated to the nine conditional factors. References [21] and [53] implemented the FR model for seismic vulnerability mapping in their research. In most cases, many of the past researches and studies used the FR model to develop landslide susceptibility models (e.g., [10,12,13,29,54,55]).…”
Section: Application Of Frequency Ratio (Fr) Modelmentioning
confidence: 99%
“…Seismic vulnerability in this study was expressed in terms of an index to analyze conditions controlled by physical and environmental factors, which increase the susceptibility of a community in the study area to the impact of seismic hazards [15]. Several methods for the seismic vulnerability mapping have been developed, proposed and adopted in recent years including Geographic Information System (GIS)-based multicriteria decision analysis (MCDA) [16], simple additive weighting (SAW) [16], analytical hierarchy process (AHP) [17][18][19], analytical network process (ANP) [17,20], logistic regression (LR) [17,21,22], support vector machine (SVM) [21], artificial neural network (ANN) [23], ANP-ANN [24], random forest (RF) along with decision tree (DT) and frequency ratio (FR) by [25] and step-wise weight assessment ratio analysis (SWARA) [26]. Reference [17] also combined various models to produce four hybrid models of; (1) fuzzy logic (fuzzy) with logistic regression (LR) (abbreviated as fuzzy-LR), (2) fuzzy with analytical network process (ANP) and AHP (abbreviated as A-fuzzy), (3) ANP and AHP with ordered weight averaging (OWA) (abbreviated as (A-OWA) and (4) OWA-LR.…”
Sabah is prone to seismic activities due to its location, being geographically located near the boundaries of three major active tectonic plates; the Eurasian, India-Australia, and Philippine-Pacific plates. The 6.0 Mw earthquake that occurred in Ranau, Sabah, on 15 June 2015 which caused 18 casualties, all of them climbers of Mount Kinabalu, raised many issues, primarily the requirements for seismic vulnerability assessment for this region. This study employed frequency ratio (FR)–index of entropy (IoE) and a combination of (FR-IoE) with an analytical hierarchy process (AHP) to map seismic vulnerability for Ranau, Sabah. The results showed that the success rate and prediction rate for the areas under the relative operating characteristic (ROC) curves were 0.853; 0.856 for the FR-IoE model and 0.863; 0.906 for (FR-IoE) AHP, respectively, with the highest performance achieved using the (FR-IoE) AHP model. The vulnerability maps produced were classified into five classes; very low, low, moderate, high, and very high seismic vulnerability. Seismic activities density ratio analysis performed on the final seismic vulnerability maps showed that high seismic activity density ratios were observed for high vulnerability zones with the values of 9.119 and 8.687 for FR-IoE and (FR-IoE) AHP models, respectively.
“…Han et al [24], in their paper entitled "Performance of Logistic Regression and Support Vector Machines for Seismic Vulnerability Assessment and Mapping: A Case Study of the 12 September 2016 ML5.8 Gyeongju Earthquake, South Korea", assess and map seismic vulnerability of the 2016 Gyeongju earthquake in Gyeongju, Korea. The LR and support vector machine (SVM) models were used to assess and map the seismic vulnerability.…”
Section: Sustainable Applications Of Rs and Gis Technologiesmentioning
The Special Issue on "Sustainable Applications of Remote Sensing and Geospatial Information Systems to Earth Observations" is published. A total of 20 qualified papers are published in this Special Issue. The topics of the papers are the application of remote sensing and geospatial information systems to Earth observations in various fields such as (1) object change detection, (2) air pollution, (3) earthquakes, (4) landslides, (5) mining, (6) biomass, (7) groundwater, and (8) urban development using the techniques of remote sensing and geospatial information systems. More than 100 researchers have participated in this Special Issue. We hope that this Special Issue is helpful for sustainable applications.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.