Search

We construct a continuous 24-h daily fine particulate matter (PM2.5)record with 2×2 km2 resolution over eastern China, South Korea, and Japan for 2011–2022 by applying a random forest (RF) algorithm to aerosol optical depth (AOD) observations from the Geostationary Ocean Color Imager (GOCI) I and II satellite instruments. The RF uses PM2.5 observations from the national surface networks as training data. PM2.5 network data starting in 2015 in South Korea are extended to pre-2015 with a RF trained on other air quality data available from the network including PM10. PM2.5 network data starting in 2014 in China are supplemented by pre-2014 data from the US embassy and consulates. Missing AODs in the GOCI data are gap-filled by a separate RF fit. We show that the resulting GOCI PM2.5 dataset is successful in reproducing the surface network observations including extreme events, and that the network data in the different countries are representative of population-weighted exposure. We find that PM2.5 peaked in 2014 (China) and 2013 (South Korea, Japan), and has been decreasing steadily since with no region left behind. We quantify the population in each country exposed to annual PM2.5 in excess of national ambient air quality standards and how this exposure evolves with time. The long record for the Seoul Metropolitan Area (SMA) shows a steady decrease from 2013 to 2022 that was not present in the first five years of AirKorea network PM2.5 measurements. Mapping of an extreme pollution event in Seoul with GOCI PM2.5 shows a predicted distribution indistinguishable from the dense urban network observations, while our previous 6×6 km2 product smoothed local features. Our product should be useful for public health studies where long-term spatial continuity of PM2.5 information is essential. [ABSTRACT FROM AUTHOR]

Procambarus clarkii is a notorious invasive species that has led to ecological concerns owing to its high viability and rapid reproduction. South Korea, a country exposed to a high risk of introduction of invasive species due to active international trade, has suffered from recent massive invasions by invasive species, necessitating the evaluation of potential areas requiring intensive monitoring. In this study, we developed two different types of species distribution models, CLIMEX and random forest, for P. clarkii using occurrence records from the United States. The potential distribution in the United States was predicted along coastal lines and inland regions located below 40°N latitude The model was then applied to evaluate the potential distribution in South Korea, and an ensemble map was constructed to identify the most vulnerable domestic regions. According to both models, the domestic potential distribution was highest in most areas located at low altitudes. In the ensemble model, most of the low-altitude western regions, the eastern coast, and some southern inland regions were predicted to be suitable for the distribution of P. clarkii, and a similar distribution pattern was predicted when the model was projected into the future climate. Through this study, it is possible to secure basic data that can be used for the early monitoring of the introduction and subsequent distribution of P. clarkii. [ABSTRACT FROM AUTHOR]

Prediction of solar energy data is very crucial for the effective utilization of freely available renewable energy abundantly in nature. Solar energy data are widely available which must be carefully prepared and arranged for modelling. In this work, typical meteorological year (TMY) data made available by the Korea institute of energy research (KIER) and the National renewable energy laboratory (NREL) are used for modelling in different phases. TMY data at single-point location and multiple locations from KIER are initially used for training of machine learning (ML) algorithms. Later, the TMY data from NREL and KIER are combined and then modelled using radius nearest neighbour (RNN), decision tree regressor (DTR), random forest regressor (RFR), and X-gradient boosting (XGB) algorithms. The solar energy parameters modelled in this work are dew point temperature (DPT), dry bulb temperature (DBT), relative humidity (RH), surface pressure (SP), windspeed (WS), and solar insolation of horizontal plane (IHP). Quantitative analysis of the algorithms is also performed in each stage of the work. The modelling indicates that the DBT, DPT, RH, and SP are able to be predicted with a minimum accuracy of over 90% in each stage. The WS and IHP data when modelled from a single-source TMY data provide superior accuracy than when they are combined. RFR and XGB have outperformed overall as they provide good accuracy for WS and IHP data as well. RNN and DTR achieved 100% accuracy in training, while RFR and XGB showed slightly lower training accuracy due to their avoidance of overfitting. There are errors in testing for RNN/DTR. Using RNN/DTR, the training errors are 0% in all cases, while in some cases like DTP the error by RFR/XGB up to 3%, whereas RNN/DTR testing errors go up to 5% and in case of RFR/XGB they are up to 7.5%. For RH modelling RFR/XGB, training errors are max 6%. RNN/DTR testing errors go up to 11%, while for RFR/XGB up to 7.5% which indicates their robustness. It is observed that many solar parameters, when combined with different source data, can be predicted easily with good accuracy, while WS and IHP become a little bit challenging to model. [ABSTRACT FROM AUTHOR]

The wheels of railway vehicles are of paramount importance in relation to railroad operations and safety. Currently, the management of railway vehicle wheels is restricted to post-event inspections of the wheels whenever physical phenomena, such as abnormal vibrations and noise, occur during the operation of railway vehicles. To address this issue, this paper proposes a method for predicting abnormalities in railway wheels in advance and enhancing the learning and prediction performance of machine learning algorithms. Data were collected during the operation of Line 4 of the Busan Metro in South Korea by directly attaching sensors to the railway vehicles. Through the analysis of key factors in the collected data, factors that can be used for tire condition classification were derived. Additionally, through data distribution analysis and correlation analysis, factors for classifying tire conditions were identified. As a result, it was determined that the z-axis of acceleration has a significant impact, and machine learning techniques such as SVM (Linear Kernel, RBF Kernel) and Random Forest were utilized based on acceleration data to classify tire conditions into in-service and defective states. The SVM (Linear Kernel) yielded the highest recognition rate at 98.70%. [ABSTRACT FROM AUTHOR]

Korea is showing the fastest trend in the world in population aging; there is a high interest in the elderly population nationwide. Among the common chronic diseases, the elderly tends to have a high incidence of depression. That said, it has been vital to focus on preventing depression in the elderly in advance. Hence, this study aims to select the factors related to depression in low-income seniors identified in previous studies and to develop a prediction model. In this study, 2975 elderly people from low-income families were extracted using the 13th-year data of the Korea Welfare Panel Study (2018). Decision trees, logistic regression, neural networks, and random forest were applied to develop a predictive model among the numerous data mining techniques. In addition, the wrapper's stepwise backward elimination, which finds the optimal model by removing the least relevant factors, was applied. The evaluation of the model was confirmed via accuracy. It was verified that the final prediction model, in the case of a decision tree, showed the highest predictive power with an accuracy of 97.3%. Second, psychological factors, leisure life satisfaction, social support, subjective health awareness, and family support ranked higher than demographic factors influencing depression. Based on the results, an approach focused on psychological support is much needed to manage depression in low-income seniors. As predicting depression in the elderly varies on numerous influencing factors, using a decision tree may be beneficial to establish a firm prediction model to identify vital factors causing depression in the elderly population. [ABSTRACT FROM AUTHOR]

Simple Summary: Efforts to prevent snakebites on hiking trails are crucial because such accidents can lead to fatalities, particularly in isolated areas. This study developed random forest species distribution models to predict the potential habitats of Gloydius spp. in South Korea, verified a significant relationship between the habitat distribution and actual snakebite accidents at the regional level, and assessed snakebite risks on trails in the national parks of Korea. The model identified high-risk snakebite areas, which will help to develop preventative solutions for hikers' safety. Snakebites can pose a significant threat to human health as the destruction of natural habitats and increased human intrusion into ecosystems result in more frequent encounters with snakes. Mitigation measures for snakebites are particularly crucial for hiking trails where transportation of snakebite victims to medical facilities is challenging due to limited emergency resources and difficult access. This study employed a random forest-based species distribution model approach to investigate the potential habitats of Gloydius spp., specifically Gloydius saxatilis, Gloydius brevicaudus, and Gloydius ussuriensis, in South Korea and to assess the snakebite risk in national parks. Potential habitats of Gloydius spp. were identified and visualized by overlaying binary maps derived from species distribution models (SDMs) of each Gloydius spp. that corresponded to high-risk snakebite areas. In addition, hiking trails with high snakebite risk in the national parks were identified after demonstrating the statistical correlation between the potential habitat distribution of Gloydius spp. and the actual snakebite incidents in major regions of South Korea. The primary environmental variables determining Gloydius spp. habitat were the topographic position index, slope, and the annual average of the maximum and minimum temperatures. The potential habitat of G. saxatilis generally appeared in high-altitude mountainous areas, mostly in the eastern part of the study area. Favorable habitats for G. brevicaudus and G. ussuriensis were predominantly located in mountainous areas throughout the study area, with the exception of some high-altitude mountainous terrain in the east. The number of snakebite incidents per 10,000 people was significantly correlated with the area ratio of Gloydius spp. potential habitat (Spearman's rho = 0.638, p < 0.01). The proportion of snakebite risk areas among national parks in South Korea ranged from 18% to 57%. This study can support practical solutions to prevent injuries and fatalities among hikers due to snakebites by identifying areas with a high risk of snakebite accidents at the hiking-trail level. [ABSTRACT FROM AUTHOR]

Authorities and policymakers in Korea have recently prioritized improving fire prevention and emergency response. Governments seek to enhance community safety for residents by constructing automated fire detection and identification systems. This study examined the efficacy of YOLOv6, a system for object identification running on an NVIDIA GPU platform, to identify fire-related items. Using metrics such as object identification speed, accuracy research, and time-sensitive real-world applications, we analyzed the influence of YOLOv6 on fire detection and identification efforts in Korea. We conducted trials using a fire dataset comprising 4000 photos collected through Google, YouTube, and other resources to evaluate the viability of YOLOv6 in fire recognition and detection tasks. According to the findings, YOLOv6's object identification performance was 0.98, with a typical recall of 0.96 and a precision of 0.83. The system achieved an MAE of 0.302%. These findings suggest that YOLOv6 is an effective technique for detecting and identifying fire-related items in photos in Korea. Multi-class object recognition using random forests, k-nearest neighbors, support vector, logistic regression, naive Bayes, and XGBoost was performed on the SFSC data to evaluate the system's capacity to identify fire-related objects. The results demonstrate that for fire-related objects, XGBoost achieved the highest object identification accuracy, with values of 0.717 and 0.767. This was followed by random forest, with values of 0.468 and 0.510. Finally, we tested YOLOv6 in a simulated fire evacuation scenario to gauge its practicality in emergencies. The results show that YOLOv6 can accurately identify fire-related items in real time within a response time of 0.66 s. Therefore, YOLOv6 is a viable option for fire detection and recognition in Korea. The XGBoost classifier provides the highest accuracy when attempting to identify objects, achieving remarkable results. Furthermore, the system accurately identifies fire-related objects while they are being detected in real-time. This makes YOLOv6 an effective tool to use in fire detection and identification initiatives. [ABSTRACT FROM AUTHOR]

Recently, various corporate failure prediction models that use machine learning techniques have received considerable attention. In particular, using a sequence of a company's historical information, rather than just the most recent information, yields better predictive performance by adopting recurrent neural networks (RNNs) and long short-term memory (LSTM) algorithms in the United States market. Similarly, we evaluate whether these results hold in emerging market contexts using listed companies in Korea. We also compare the logistic regression, random forest, RNN, LSTM, and an ensemble model combining these four techniques. The random forest model with recent information outperforms the other models, indicating that corporate failure prediction models for immature markets, unlike those for developed markets, might have to focus more on recent information rather than on the historical sequence of corporate performance. [ABSTRACT FROM AUTHOR]

Background: The present study aimed to evaluate the performance of several machine learning (ML) algorithms in predicting 1‐year afatinib continuation and 2‐year survival after afatinib initiation and to identify the differences in survival outcomes between ML‐classified strata. Methods: Data that were also used in the RESET study were retrospectively collected from 16 hospitals in South Korea. A stratified random sampling method was applied to split the data into training and test sets (70:30 split ratio). Clinical information, such as age, sex, tumor stage, smoking, performance status, metastasis, type of metastasis, dose adjustment, and pathologic information on EGFR mutations were inputted. Training was performed using eight ML algorithms: logistic regression, decision tree, deep neural network, random forest, support vector machine, boosting, bagging, and the naïve Bayes classifier. The model performance was assessed based on sensitivity, specificity, and accuracy. Area under the receiver operator characteristic curve (AUC) was calculated and compared between the ML models using DeLong's test. A Kaplan–Meier (KM) curve was used to visualize the identified strata obtained from the ML models. Results: No significant differences in the input variables were observed between the training and test datasets. The best‐performing models were support vector machine in predicting 1‐year afatinib continuation (AUC 0.626) and decision tree in 2‐year survival after afatinib start (AUC 0.644), although the performances of the ML models were comparable and did not display any predictive roles. KM analysis and log‐rank test revealed significant differences between the strata identified from the ML model (p < 0.001) in terms of both time‐on‐treatment (TOT) and overall survival (OS). Conclusion: The performances of ML models in our study found no discernible roles in predicting afatinib‐related outcomes, although the identified strata revealed different TOT and OS in the KM analysis. This implies the strength of ML in predicting the survival outcome, as well as the limitation of electronic medical record‐based variables in ML algorithms. Careful consideration of variable inclusion is likely to improve the general model performance. [ABSTRACT FROM AUTHOR]

In this study, we aimed to develop a prediction model of the solid solutions concentration (SSC) and moisture content (MC) in oriental melon with snapshot-type hyperspectral imagery (Visible (VIS): 460–600 nm, 16 bands; Red-Near infrared (Red-NIR): 600–860 nm, 15 bands) using a machine learning model. The oriental melons were cultivated in a hydroponic greenhouse, Republic of Korea, and a total of 91 oriental melons that were harvested from March to April of 2022 were used as samples. The SSC and MC of the oriental melons were measured using destructive methods after taking hyperspectral imagery of the oriental melons. The reflectance spectrum obtained from the hyperspectral imagery was processed by the standard normal variate (SNV) method. Variable importance in projection (VIP) scores were used to select the bands related to SSC and MC. As a result, ten (609, 736, 561, 849, 818, 489, 754, 526, 683, and 597 nm) and six (609, 736, 561, 818, 849, and 489 nm) bands were selected for the SSC and MC, respectively. Four machine learning models, support vector regression (SVR), ridge regression (RR), K-nearest neighbors regression (K-NNR), and random forest regression (RFR), were used to develop models to predict SSC and MC, and their performances were compared. The SVR showed the best performance for predicting both the SSC and MC of the oriental melons. The SVR model achieved a relatively high accuracy with R2 values of 0.86 and 0.74 and RMSE values of 1.06 and 1.05 for SSC and MC, respectively. However, it will be necessary to carry out more experiments under various conditions, such as differing maturities of fruits and varying light sources and environments, to achieve more comprehensive predictions and apply them to monitoring robots in the future. Nevertheless, it is considered that the snapshot-type hyperspectral imagery aided by SVR would be a useful tool to predict the SSC and MC of oriental melon. In addition, if the maturity classification model for the oriental melon can be applied to fields, it could lead to less labor and result in high-quality oriental melon production. [ABSTRACT FROM AUTHOR]

Highly pathogenic avian influenza (HPAI) is a fatal zoonotic disease that damages the poultry industry and endangers human lives via exposure to the pathogen. A risk assessment model that precisely predicts high‐risk groups and occurrence of HPAI infection is essential for effective biosecurity measures that minimize the socio‐economic losses of massive outbreaks. However, the conventional risk prediction approaches have difficulty incorporating the broad range of factors associated with HPAI infections at poultry holdings. Therefore, it is difficult to accommodate the complexity of the dynamic transmission mechanisms and generate risk estimation on a real‐time basis. We proposed a continuous risk prediction framework for HPAI occurrences that used machine learning algorithms (MLAs). This integrated environmental, on‐farm biosecurity, meteorological, vehicle movement tracking, and HPAI wild bird surveillance data to improve accuracy and timeliness. This framework consisted of (i) the generation of 1788 predictors from six types of data and reconstructed them with an outcome variable into a data mart based on a temporal assumption (i.e. infected period and day‐ahead forecasting); (ii) training of the predictors with the temporally rearranged outcome variable that corresponded to HPAI H5N6 infected state at each individual farm on daily basis during the 2016–2017 HPAI epidemic using three different MLAs [Random Forest, Gradient Boosting Machine (GBM), and eXtreme Gradient Boosting]; (iii) predicting the daily risk of HPAI infection during the 2017–2018 HPAI epidemic using the pre‐trained MLA models for each farm across the country. The models predicted the high risk to 8–10 out of 19 infected premises during the infected period in advance. The GBM MLAs outperformed the 7‐day forecasting of HPAI prediction at individual poultry holdings, with an area under the curve (AUC) of receiver operating characteristic of 0.88. Therefore, this approach enhances the flexibility and timing of interventions against HPAI outbreaks at poultry farms. [ABSTRACT FROM AUTHOR]

A two-dimensional visibility estimation model was developed, based on random forest (RF), a machine learning-based technique. A geostatistical method was introduced into the visibility estimation model for the first time to interpolate point measurement data to gridded data spatially with a pixel size of 10 km. The RF-based model was trained using gridded visibility data, as well as meteorological and air pollution input variable data, for each location in South Korea, which were characterized by complex geographical features and high air pollution levels. Generally, relative humidity was the most important input variable for the visibility estimation (average mean decrease accuracy: 35%). However, PM2.5 tended to be the most crucial variable in polluted regions. The spatial interpolation was found to result in an additional visibility estimation error of 500 m in locations where no adjacent visibility observations within 0.2° were available. The performance of the proposed model was preliminarily assessed. Generally, the best detection performance was achieved in good visibility conditions (visibility range: 10 to 20 km). This study is the first to demonstrate a visibility estimation model based on a geostatistical method and machine learning, which can provide visibility information in locations for which no observations exist. [ABSTRACT FROM AUTHOR]

Forecasting wind speed near the surface with high-spatial resolution is beneficial in agricultural management. There is a discrepancy between the wind speed information required for agricultural management and that produced by weather agencies. To improve crop yield and increase farmers' incomes, wind speed prediction systems must be developed that are customized for agricultural needs. The current study developed a high-resolution wind speed forecast system for agricultural purposes in South Korea. The system produces a wind speed forecast at 3 m aboveground with 100-m spatial resolution across South Korea. Logarithmic wind profile, power law, random forests, support vector regression, and extreme learning machine were tested as candidate methods for the downscaling wind speed data. The wind speed forecast system developed in this study provides good performance, particularly in inland areas. The machine learning–based methods give the better performance than traditional methods for downscaling wind speed data. Overall, the random forests are considered the best downscaling method in this study. Root mean square error and mean absolute error of wind speed prediction for 48 h using random forests are approximately 0.8 m/s and 0.5 m/s, respectively. [ABSTRACT FROM AUTHOR]

In predicting energy consumption, classic econometric and statistical models are used to forecast energy consumption. These models may have limitations in an increasingly fast-changing energy market, which requires big data analysis of energy consumption patterns and relevant variables using complex mathematical tools. In current literature, there are minimal comparison studies reviewing machine learning algorithms to predict energy consumption in Korea. To bridge this gap, this paper compared three different machine learning algorithms, namely the Random Forest (RF) model, XGBoost (XGB) model, and Long Short-Term Memory (LSTM) model. These algorithms were applied in Period 1 (prior to the onset of the COVID-19 pandemic) and Period 2 (after the onset of the COVID-19 pandemic). Period 1 was characterized by an upward trend in energy consumption, while Period 2 showed a reduction in energy consumption. LSTM performed best in its prediction power specifically in Period 1, and RF outperformed the other models in Period 2. Findings, therefore, suggested the applicability of machine learning to forecast energy consumption and also demonstrated that traditional econometric approaches may outperform machine learning when there is less unknown irregularity in the time series, but machine learning can work better with unexpected irregular time series data. [ABSTRACT FROM AUTHOR]

This paper presents two data-driven virtual sensors to estimate the time-series of the probability of high-concentration occurrence of naturally occurring radioactive materials (NORMs; 238U and 222Rn) in groundwater based on the in-situ groundwater quality monitoring data and geological information. The random forest was applied to estimate the NORM concentration based on the actual in-situ groundwater quality data, rock type, and the aquifer depth. Additionally, this study proposes three data sampling techniques (i.e., under-sampling, synthetic minority over-sampling, and a complex sampling) to improve the model applicability and accuracy. The developed models were validated using the actual data acquired from 201 locations in South Korea. The models for 238U and 222Rn showed estimation accuracies of 85% and 80%, respectively; the models with over-sampling showed better performance. All the results verified the usefulness of the developed models as virtual sensors for providing immediate information on the in-situ presence of NORMs in groundwater. [Display omitted] • Naturally occurring radioactive material (NORMs) in groundwater can be problematic. • Virtual sensor for in-situ estimating the probability of high-concentration of NORMs. • Major geochemical factors for NORMs were evaluated using statistical analyses. • Methods for solving data imbalance problem in training virtual sensor were proposed. • Major factors for the estimation were consistent with the statistical analysis result. [ABSTRACT FROM AUTHOR]

This study investigated how changes in weather factors affect the prevalence of conjunctivitis using public big data in South Korea. A total of 1,428 public big data entries from January 2013 to December 2019 were collected. Disease data and basic climate/air pollutant concentration records were collected from nationally provided big data. Meteorological factors affecting eye diseases were identified using multiple linear regression and machine learning analysis methods such as extreme gradient boosting (XGBoost), decision tree, and random forest. The prediction model with the best performance was XGBoost (1.180), followed by multiple regression (1.195), random forest (1.206), and decision tree (1.544) when using root mean square error (RMSE) values. With the XGBoost model, province was the most important variable (0.352), followed by month (0.289) and carbon monoxide exposure (0.133). Other air pollutants including sulfur dioxide, PM10, nitrogen dioxides, and ozone showed low associations with conjunctivitis. We identified factors associated with conjunctivitis using traditional multiple regression analysis and machine learning techniques. Regional factors were important for the prevalence of conjunctivitis as well as the atmosphere and air quality factors. [ABSTRACT FROM AUTHOR]

Machine learning (ML) and deep neural network (DNN) techniques are promising tools. These can advance mathematical crop modelling methodologies that can integrate these schemes into a process-based crop model capable of reproducing or simulating crop growth. In this study, an innovative hybrid approach for estimating the leaf area index (LAI) of paddy rice using climate data was developed using ML and DNN regression methodologies. First, we investigated suitable ML regressors to explore the LAI estimation of rice based on the relationship between the LAI and three climate factors in two administrative rice-growing regions of South Korea. We found that of the 10 ML regressors explored, the random forest regressor was the most effective LAI estimator, and it even outperformed the DNN regressor, with model efficiencies of 0.88 in Cheorwon and 0.82 in Paju. In addition, we demonstrated that it would be feasible to simulate the LAI using climate factors based on the integration of the ML and DNN regressors in a process-based crop model. Therefore, we assume that the advancements presented in this study can enhance crop growth and productivity monitoring practices by incorporating a crop model with ML and DNN plans. [ABSTRACT FROM AUTHOR]

This study was a multicenter retrospective cohort study of term nulliparous women who underwent labor, and was conducted to develop an automated machine learning model for prediction of emergent cesarean section (CS) before onset of labor. Nine machine learning methods of logistic regression, random forest, Support Vector Machine (SVM), gradient boosting, extreme gradient boosting (XGBoost), light gradient boosting machine (LGBM), k-nearest neighbors (KNN), Voting, and Stacking were applied and compared for prediction of emergent CS during active labor. External validation was performed using a nationwide multicenter dataset for Korean fetal growth. A total of 6549 term nulliparous women was included in the analysis, and the emergent CS rate was 16.1%. The C-statistics values for KNN, Voting, XGBoost, Stacking, gradient boosting, random forest, LGBM, logistic regression, and SVM were 0.6, 0.69, 0.64, 0.59, 0.66, 0.68, 0.68, 0.7, and 0.69, respectively. The logistic regression model showed the best predictive performance with an accuracy of 0.78. The machine learning model identified nine significant variables of maternal age, height, weight at pre-pregnancy, pregnancy-associated hypertension, gestational age, and fetal sonographic findings. The C-statistic value for the logistic regression machine learning model in the external validation set (1391 term nulliparous women) was 0.69, with an overall accuracy of 0.68, a specificity of 0.83, and a sensitivity of 0.41. Machine learning algorithms with clinical and sonographic parameters at near term could be useful tools to predict individual risk of emergent CS during active labor in nulliparous women. [ABSTRACT FROM AUTHOR]

We use 2011–2019 aerosol optical depth (AOD) observations from the Geostationary Ocean Color Imager (GOCI) instrument over East Asia to infer 24 h daily surface fine particulate matter (PM 2.5) concentrations at a continuous 6 × 6 km 2 resolution over eastern China, South Korea, and Japan. This is done with a random forest (RF) algorithm applied to the gap-filled GOCI AODs and other data, including information encoded in GOCI AOD retrieval failure and trained with PM 2.5 observations from the three national networks. The predicted 24 h GOCI PM 2.5 concentrations for sites entirely withheld from training in a 10-fold cross-validation procedure correlate highly with network observations (R2 = 0.89) with a single-value precision of 26 %–32 %, depending on the country. Prediction of the annual mean values has R2 = 0.96 and a single-value precision of 12 %. GOCI PM 2.5 is only moderately successful for diagnosing local exceedances of the National Ambient Air Quality Standard (NAAQS) because these exceedances are typically within the single-value precisions of the RF and also because of RF smoothing of extreme PM 2.5 concentrations. The area-weighted and population-weighted trends of GOCI PM 2.5 concentrations for eastern China, South Korea, and Japan show steady 2015–2019 declines consistent with surface networks, but the surface networks in eastern China and South Korea underestimate population exposure. Further examination of GOCI PM 2.5 fields for South Korea identifies hot spots where surface network sites were initially lacking and shows 2015–2019 PM 2.5 decreases across the country, except for flat concentrations in the Seoul metropolitan area. Inspection of the monthly PM 2.5 time series in Beijing, Seoul, and Tokyo shows that the RF algorithm successfully captures observed seasonal variations in PM 2.5 , even though AOD and PM 2.5 often have opposite seasonalities. The application of the RF algorithm to urban pollution episodes in Seoul and Beijing demonstrates high skill in reproducing the observed day-to-day variations in air quality and spatial patterns on the 6 km scale. A comparison to a Community Multiscale Air Quality (CMAQ) simulation for the Korean peninsula demonstrates the value of the continuous GOCI PM 2.5 fields for testing air quality models, including over North Korea, where they offer a unique resource. [ABSTRACT FROM AUTHOR]

This study aimed to evaluate the effectiveness of wireless emergency alerts (WEAs) on social distancing policy. The Republic of Korea has been providing information to the public through WEAs using mobile phones. This study used five data sets: WEA messages, news articles including the keyword "COVID-19," the number of confirmed COVID-19 patients, public foot traffic data, and the government's social distancing level. The WEAs were classified into two topics—"warning" and "guidance"—using a random forest model. The results of the correlation analysis and further detailed analysis confirmed that the "warning" WEA topic and number of news articles significantly affected public foot traffic. However, the "guidance" topic was not significantly associated with public foot traffic. In general, the Korean government's WEAs were effective at encouraging the public to follow social distance recommendations during the COVID-19 pandemic. In particular, the "warning" WEA topic, by providing information about the relative risk directly concerning the recipients, was significantly more effective than the "guidance" topic. [ABSTRACT FROM AUTHOR]

[Display omitted] • Developed the AquaShan Index, an advanced method for river health assessment combining biological and environmental data. • Applied Random Forest analysis for precise weighting of environmental variables in river ecosystem evaluation. • Analyzed extensive data from South Korea's river ecosystems, focusing on benthic macroinvertebrates and environmental factors. • Demonstrated greater accuracy of the AquaShan Index in identifying river health issues compared to traditional indices. • Provided a practical, adaptable tool for informed river management and conservation decision-making. This study introduces the AquaShan Index, an innovative approach to assessing river health, which integrates biological indicators with key environmental variables. This index represents an advancement over traditional methods, like the Shannon Diversity Index, by including both species diversity and environmental factors such as dissolved oxygen, pH, and water temperature. Utilizing data from the Stream/River Ecosystem Survey and Health Assessment in South Korea, covering 5,952 data over three years, the study combines species data of benthic macroinvertebrates with environmental parameters, applying Random Forest analysis to weight these variables. The AquaShan Index formula incorporates species' pollution tolerance scores, offering a nuanced perspective on ecosystem resilience to environmental stressors. Our results show a wide range of river health conditions, with the AquaShan Index demonstrating greater sensitivity and specificity in identifying ecological impairments compared to traditional indices. The inclusion of pollution tolerance scores and environmental variables provides a more comprehensive assessment of river health, revealing a higher number of sites with ecological challenges than previously identified. The AquaShan Index's adaptability and comprehensive approach make it a valuable tool for informed decision-making in river conservation and management, offering an enhanced methodology for assessing and sustaining river ecosystem health. [ABSTRACT FROM AUTHOR]

PM 10 and SO 2 are major air pollutants in Ulsan, the largest industrial city in South Korea. The purpose of this study was to evaluate the effectiveness of pollution reduction policies by comparing the pollution characteristics of PM 10 and SO 2 between 2012 and 2020. The concentrations of PM 10 and SO 2 have substantially decreased over time. Local industry was the primary driver of SO 2 levels, whereas various sources were associated with PM 10 levels. In 2012 and 2020, the spatial distributions of SO 2 were similar, with hot spots around petrochemical and nonferrous industrial complexes, whereas the sources of PM 10 differed between the two years. Although industrial emissions of PM 10 decreased from 2012 to 2020, the concentrations derived from an air dispersion model were slightly higher in 2020 due to changes in both wind patterns and the strength of the emission sources. In contrast, modeled SO 2 levels were higher in 2012 than in 2020 due to greater emissions in 2012. Eastern and northeastern China were identified as potential source areas for the long-range atmospheric transport of both PM 10 and SO 2 , though its influence was significantly lower in 2020. A comparison of the PM 10 and SO 2 concentrations predicted by a random forest model and the measured concentrations indicates that government policies have been effective in reducing the emissions and ambient air concentrations of both pollutants. In conclusion, this study demonstrates that local industry remains a significant contributor to air pollution in Ulsan despite the effects of governmental policy. Furthermore, as the influence of external sources weakens, there is a pressing need to strengthen the management of local sources to effectively mitigate air pollution in Ulsan. [Display omitted] • The effectiveness of pollution reduction policies for PM 10 and SO 2 was evaluated. • The influence of LRAT and local sources has significantly decreased recently. • Governmental policies have led to a decrease in PM 10 and SO 2 concentrations. • However, local industry remains a significant contributor to air pollution. • The result of the study can be used to develop industry-specific management policies. • To 5 bullet points (maximum 85 characters, including spaces, per bullet point). [ABSTRACT FROM AUTHOR]

We use 2011-2019 aerosol optical depth (AOD) observations from the Geostationary Ocean Color Imager (GOCI) instrument over East Asia to infer 24-h daily surface fine particulate matter (PM2.5) concentrations at continuous 6x6 km² resolution over eastern China, South Korea, and Japan. This is done with a random forest (RF) algorithm applied to the gap-filled GOCI AODs and other data and trained with PM2.5 observations from the three national networks. The predicted 24-h PM2.5 concentrations for sites entirely withheld from training in a ten-fold crossvalidation procedure correlate highly with network observations (R² = 0.89) with single-value precision of 26-32% depending on country. Prediction of annual mean values has R² = 0.96 and single-value precision of 12%. The RF algorithm is only moderately successful for diagnosing local exceedances of the National Ambient Air Quality Standard (NAAQS) because these exceedances are typically within the single-value precisions of the RF, and also because of RF smoothing of extreme PM2.5 concentrations. The area-weighted and population-weighted trends of RF PM2.5 concentrations for eastern China, South Korea, and Japan show steady 2015-2019 declines consistent with surface networks, but the surface networks in eastern China and South Korea underestimate population exposure. Further examination of RF PM2.5 fields for South Korea identifies hotspots where surface network sites were initially lacking and shows 2015-2019 PM2.5 decreases across the country except for flat concentrations in the Seoul metropolitan area. Inspection of monthly PM2.5 time series in Beijing, Seoul, and Tokyo shows that the RF algorithm successfully captures observed seasonal variations of PM2.5 even though AOD and PM2.5 often have opposite seasonalities. Application of the RF algorithm to urban pollution episodes in Seoul and Beijing demonstrates high skill in reproducing the observed day-to-day variations in air quality as well as spatial patterns on the 6 km scale. Comparison to a CMAQ simulation for the Korean peninsula demonstrates the value of the continuous RF PM2.5 fields for testing air quality models, including over North Korea where they offer a unique resource. [ABSTRACT FROM AUTHOR]

Background: Alcohol use disorder (AUD) is a chronic disease with a higher recurrence rate than that of other mental illnesses. Moreover, it requires continuous outpatient treatment for the patient to maintain abstinence. However, with a low probability of these patients to continue outpatient treatment, predicting and managing patients who might discontinue treatment becomes necessary. Accordingly, we developed a machine learning (ML) algorithm to predict which the risk of patients dropping out of outpatient treatment schemes. Methods: A total of 839 patients were selected out of 2,206 patients admitted for AUD in three hospitals under the Catholic Central Medical Center in Korea. We implemented six ML models—logistic regression, support vector machine, k-nearest neighbor, random forest, neural network, and AdaBoost—and compared the prediction performances thereof. Results: Among the six models, AdaBoost was selected as the final model for recommended use owing to its area under the receiver operating characteristic curve (AUROC) of 0.72. The four variables affecting the prediction based on feature importance were the length of hospitalization, age, residential area, and diabetes. Conclusion: An ML algorithm was developed herein to predict the risk of patients with AUD in Korea discontinuing outpatient treatment. By testing and validating various machine learning models, we determined the best performing model, AdaBoost, as the final model for recommended use. Using this model, clinicians can manage patients with high risks of discontinuing treatment and establish patient-specific treatment strategies. Therefore, our model can potentially enable patients with AUD to successfully complete their treatments by identifying them before they can drop out. [ABSTRACT FROM AUTHOR]

Background & Aims: There are currently several prediction models for hepatocellular carcinoma (HCC) in chronic hepatitis B (CHB) receiving oral antiviral therapy. However, most models are based on pre‐treatment clinical parameters. The current study aimed to develop a novel and practical prediction model for HCC by using both pre‐ and post‐treatment parameters in this population. Methods: We included two treatment‐naïve CHB cohorts who were initiated on oral antiviral therapies: the derivation cohort (n = 1480, Korea prospective SAINT cohort) and the validation cohort (n = 426, the US retrospective Stanford Bay cohort). We employed logistic regression, decision tree, lasso regression, support vector machine and random forest algorithms to develop the HCC prediction model and selected the most optimal method. Results: We evaluated both pre‐treatment and the 12‐month clinical parameters on‐treatment and found the 12‐month on‐treatment values to have superior HCC prediction performance. The lasso logistic regression algorithm using the presence of cirrhosis at baseline and alpha‐foetoprotein and platelet at 12 months showed the best performance (AUROC = 0.843 in the derivation cohort. The model performed well in the external validation cohort (AUROC = 0.844) and better than other existing prediction models including the APA, PAGE‐B and GAG models (AUROC = 0.769 to 0.818). Conclusions: We provided a simple‐to‐use HCC prediction model based on presence of cirrhosis at baseline and two objective laboratory markers (AFP and platelets) measured 12 months after antiviral initiation. The model is highly accurate with excellent validation in an external cohort from a different country (AUROC 0.844) (Clinical trial number: KCT0003487). [ABSTRACT FROM AUTHOR]

The aim of this study is to evaluate and compare the performances of 5 machine learning (ML) techniques for predicting the spatial probability of landslide at Atsuma, Japan, and Mt. Umyeon, Korea. 5 ML models used are Naïve Bayes (NB), ensemble learning (random forest (RF) and adaboost (AB)), and deep learning (multilayer perceptron (MLP) and convolutional neural network (CNN)) models. Real landslide events at the study areas are randomly separated to the training set for landslide mapping and the validation set for assessing performance. To assess the performance of the used models, the resulting models are validated using receiver operating characteristic (ROC) curve. The success rate curves show that the areas under the curve (AUC) for the NB, RF, AB, MLP, and CNN are 85.1, 88.8, 88.6, 87.5, and 95.0%, respectively, at Atsuma and 68.7, 85.6, 90.5, 81.6, and 92.0%, respectively, at Mt. Umyeon. Similarly, the validation results show that the areas under the curve for the NB, RF, AB, MLP, and CNN are 84.3, 87.1, 87.1, 86.7, and 89.7%, respectively, at Atsuma and 64.9, 85.5, 83.9, 84.7, and 90.5%, respectively, at Mt. Umyeon. In addition, statistical tests such as Chi-square test and difference of proportions test show that all classified landslide susceptibility maps have statistical significance and the significant difference in classified landslide susceptibility maps from different ML models. The comparison results among 5 ML models show that the CNN model had the best performance and NB model had the worst performance in both study areas. [ABSTRACT FROM AUTHOR]

• This study assesses the traffic forecasting accuracy of the Korean PFS. • Project manager and forecaster traits minimally affected traffic forecasting errors. • Political influence had a minimal effect on traffic forecasting errors. • PFS enhances the accuracy of traffic forecasts. • This study examines policy to increase traffic forecasting accuracy. The financial crisis that beset the Republic of Korea in late 1997 had a tremendous impact on the nation's economy. In an effort to resolve the fundamental causes of this crisis, one fiscal reform, executed with the goal of improving the efficacy of government investment, was the introduction of the Preliminary Feasibility Study (PFS). PFS was initiated to determine which projects to fund in a transparent and fair way through preliminarily examinations and evaluation of the feasibility of large-scale government projects. This study assesses the achievements of PFS by investigating improvements in the accuracy of traffic forecasting results and analyzing the determinants of traffic forecasting errors in PFS. The traffic forecasting data include 42 road projects for which PFS were conducted and 113 road projects that were not subject to PFS and for which forecasts and actual out-turns were available. The results reveal that 41.3% of traffic forecasts were overestimated before the introduction of PFS. This statistic contrasts with the traffic forecasting error rate of 14.53% since PFS became mandatory. Moreover, traffic forecasts of PFS projects were also unbiased. By analyzing the determinants of traffic forecasting errors in PFS projects using the random forest approach, we also found that the characteristics of the project manager and forecaster, and the influence of politicians or political groups, had minimal effect on traffic forecasting errors. We conclude that PFS enhances the accuracy of traffic forecasts and prevents the initiation of inefficient transportation projects. Our study fills a gap in the existing literature because few studies have focused on the various factors affecting traffic forecasting error based on data and econometrics. This study also examines the effects of policy measures to mitigate overestimation and increase traffic forecasting accuracy. Our results emphasize the significant role of analysts, the need for credible guidelines and an independent evaluation institution, and the importance of a robust governance system in public investment management. [ABSTRACT FROM AUTHOR]

• A two-stage data mining method is used for identifying factors affecting the safety of taxi drivers. • Random forest analyses are adopted for the data mining. • An in-depth questionnaire survey is conducted to obtain intrinsic characteristics underlying the cause of taxi crashes. • A set of policy implications is established for enhancing taxi safety. Introduction: Traffic safety issues associated with taxis are important because the frequency of taxi crashes is significantly higher than that of other vehicle types. The purpose of this study is to derive safety implications to be used for developing policies to enhance taxi safety based on analyzing intrinsic characteristics underlying the cause of traffic accidents. Method: An in-depth questionnaire survey was conducted to collect a set of useful data representing the intrinsic characteristics. A total of 781 corporate taxi drivers participated in the survey in Korea. The proposed analysis methodology consists of two-stage data mining techniques, including a random forest method, with data that represents the working condition and welfare environment of taxi drivers. In the first stage, the drivers' intrinsic characteristics were derived to classify four types of taxi drivers: unspecified normal, work-life balanced, overstressed, and work-oriented. Next, priority was determined for classifying high-risk taxi drivers based on factors derived from the first analysis. Results: The derived policies can be categorized into three groups: 'the development of new policies,' 'the improvement of existing policies,' and 'the elimination of negative factors.' Establishing a driving capability evaluation system for elderly drivers, developing mental health management programs for taxi drivers, and inspecting the taxi's internal conditions were proposed as new policies. Improving the driver's wage system, supporting the improvement of rest facilities, and supporting the installation of security devices for protecting taxi drivers are methods for improving existing policies to reinforce the traffic safety of taxi drivers. Last, restricting overtime work for taxi drivers was proposed as a policy to eliminate negative factors for improving taxi traffic safety. Practical Applications: It is expected that by devising effective policies using the policy implications suggested in this study, taxi traffic accidents can be prevented and the quality of life of taxi drivers can be improved. [ABSTRACT FROM AUTHOR]

The Geostationary Ocean Color Imager II (GOCI-II) deployed on board GEO-KOMPSAT 2B (GK2B) satellite can observe the earth with high frequency, which is of great significance to the dynamic monitoring of regional aerosol optical depth (AOD). However, the lack of short-wave infrared bands makes the classical methods based on radiative transfer analysis difficult to achieve retrieval of aerosols. The data-driven methods based on learning from massive samples can achieve high-accuracy retrieval of aerosols on the local, regional and global scale. However, the samples required for the data-driven method are difficult to obtain due to the short operation time of the GK2B satellite and insufficient ground-based measurements. Aiming to solve such a problem, the Simplified and Robust Surface Reflectance Estimation and Simplified Aerosol Retrieval Algorithm (SEMARA) were aggregated to be applied in the study. The SEMARA method using ground-based observation can retrieve local AOD with high reliability and accuracy, which can be used to construct high-quality and high-quantity samples for training with higher efficiency compared with traditional sample construct methods. Four data-driven methods: Fully Connected Neural Network (FCNN), Random Forest (RF), Support Vector Machine (SVM) and eXtreme Gradient Boosting (XGBoost), were chosen to retrieve the AOD over South Korea in 2021 and 2022 based on GOCI-II. The results were validated and compared using ground-based measurements from the Aerosol Robotic Network (AENRONET) sites and show that the FCNN method has high accuracy and stability, followed by, the RF and XGBoost methods, while the SVM results significantly overestimate the spatial distribution of the AOD. • High-precision AOD retrieval methods with high time frequency based GOCI-II over South Korea were proposed. • A sample construction method based on SEMARA was proposed for four data-driven models. • The retrieval accuracy and reliability of these four AOD retrieval methods were verified and compared. • The FCNN has high accuracy and stability, followed by, the RF and XGBoost, while the SVM shows clear overestimation. [ABSTRACT FROM AUTHOR]

Suicide is a serious social and public health problem. Social stigma and prejudice reduce the accessibility of mental health care services for high-risk groups, resulting in them not receiving interventions and committing suicide. A suicide prediction model is necessary to identify high-risk groups in the general population. We used national medical check-up data from 2009 to 2015 in Korea. The latest medical check-up data for each subject was set as an index point. Analysis was undertaken for an overall follow-up period (index point to the final tracking period) as well as for a one-year follow-up period. The training set was cross-validated fivefold. The predictive model was trained using a random forest algorithm, and its performance was measured using a separate test set not included in the training. The analysis covered 372,813 individuals, with an average (SD) overall follow-up duration of 1.52 (1.52) years. When we predicted suicide during the overall follow-up period, the area under the receiver operating characteristic curve (AUC) was 0.849, sensitivity was 0.817, and specificity was 0.754. The performance of the predicted suicide risk model for one year from the index point was AUC 0.818, sensitivity 0.788, and specificity 0.657. This is probably the first suicide predictive model using machine learning based on medical check-up data from the general population. It could be used to screen high-risk suicidal groups from the population through routine medical check-ups. Future studies may test preventive interventions such as exercise and alcohol in these high-risk groups. [ABSTRACT FROM AUTHOR]

In general, sea farms are classified into two types depending on their locations, seawater depths, and cultivation species: underwater and above water farms. This study compared the machine learning techniques (random forest, a supervised learning technique, and K-means, an unsupervised learning technique) for mapping both underwater and above water farms using the high-resolution satellite image acquired in the sea farming areas of the South Sea of South Korea through the following steps. First, each machine learning algorithm was separately used in the given high-resolution satellite image to generate a sea surface map. Then underwater and above water farms were detected from each sea surface map. Finally, the accuracy of both the underwater and above water farms detected from the different sea surface maps was assessed. The experimental results led to the following conclusions. First, random forest, a supervised learning technique, has better performance than K-means, an unsupervised learning technique, for detecting both underwater and above water farms from the given high-resolution satellite image. Second, only a few misclassification errors occurred in both the underwater and above water farms detected by the random forest algorithm in the given image, with the underwater farms misclassified as having above water features, or vice versa. [ABSTRACT FROM AUTHOR]

• Machine learning (ML) was applied to model tree mortality and growth reduction. • Biogeochemical model produced ecophysiological variables for the ML modeling. • Random forest (RF) showed the best performance among the MLs. • Sugar and starch contents were identified as key parameters in the ML models. Massive tree diebacks and abnormal growth reduction have been more reported across continents, which calls an attention for methods that can assess the physiological stress of tree growth. This study aims to develop machine learning models (MLM) to explain tree mortality and growth reduction in terms of the key parameters of climate and tree ecophysiology in temperate forests of Korea. For this, we produced various ecophysiological parameters with a process-based vegetation model that was used as inputs to multiple machine learning algorithms and the conventional multiple linear regression (MLR). As a result, the MLMs outperformed the MLR. Among the MLMs, random forest (RF) showed the highest overall accuracy and AUC for both tree mortality (79% and 0.84±0.09, respectively) and growth reduction (61% and 0.63 ± 0.09). Winter temperature along with previous-year autumn soluble sugar content and precipitation were the most important in determining the tree mortality, while the growth reduction was largely regulated by current-year conditions such spring and autumn precipitation and summer starch content. Also, in the proportion of variance (communalities) of PCA, the precipitation and soluble sugar showed low values 0.15 and 0.29, respectively, which low communalities (<0.3) indicate that these variables have limited shared variance with other variables in the analysis. This study indicates the combined use of vegetation model and MLM can give reciprocal benefits to each other by providing ecophysiological parameters for MLM, otherwise hard to get, and coping with convoluted process of tree vitality undescribed yet in the vegetation model, respectively. [ABSTRACT FROM AUTHOR]

Members of the armed forces rely on having an effective and efficient medical evacuation (MEDEVAC) process for evacuating casualties from the battlefield to medical treatment facilities during combat operations. Determining which MEDEVAC units to task to respond to incoming requests for service, known as the MEDEVAC dispatching problem, is a natural sequential decision making problem and is the focus of this paper. Unlike previous research in this area, this paper incorporates triage classification errors and the possibility of having blood transfusion kits on board select MEDEVAC units. A discounted, infinite-horizon continuous-time Markov decision process model is formulated to examine the MEDEVAC dispatching problem and to compare generated dispatching policies to the currently practiced United States military policy. We utilize an approximate dynamic programming (ADP) technique that leverages a random forest value function approximation within an approximate policy iteration algorithmic framework to develop high-quality policies for both small-scale and large-scale problem instances. A representative planning scenario involving joint combat operations in South Korea is developed and utilized to investigate the differences between the various policies. Results from the analysis indicate that applying ADP techniques can improve current practices by as much as 29% with regard to a life-saving performance metric. This research is of particular interest to the military medical community and can inform the procedures of future military MEDEVAC operations. [ABSTRACT FROM AUTHOR]

Stand productivity research has mainly focused on increasing yield and has recently begun to examine changes in carbon storage. The Korean government is interested in finding ways to increase forest carbon capture to meet carbon neutrality requirements because approximately 63% of the land is covered by forests. In addition, 69% of these forests are older than 30 years old and their productivity and aboveground carbon storage rates are expected to decline. The purpose of this study was to examine the effect of quadratic mean diameter (QMD), stand basal area, site index, slope, climate (MAT and MAP), stand age, stand structural diversity, and stand composition on the productivity of aging Korean red pine-dominated stands. Based on the effects of these factors, we explored how to manage pine forests with the focus of increasing their productivity. Random forest regression was used for the analysis, and periodic basal area increment (PBAI) was used as the dependent variable of stand productivity. Our results show that the most influential factor on stand productivity was QMD. PBAI dramatically decreased from approximately 0.8 to 0.53 m2/ha·year as QMD increased up to 18 cm. Since diameter (QMD) increment is closely associated with changes in tree competition; increasing tree competition with increasing QMD and stand basal area may lead to decreases in PBAI owing to decreases in growth rate due to space and resource limitations and increases in mortality. PBAI decreased when basal area increased from 22 to 51.5 m2/ha. PBAI increased for site index values between 8 and 12.5 m and decreased for stand age values up to approximately 31 years. For climate factors, PBAI generally increased with increasing MAP and slightly increased as MAT increased up to approximately 11.2 °C and then decreased at higher MAT. PBAI initially increased with increasing slope values, decreased with values lower than 15°, and remained stable at slope values in the range of 16–34°. Stand structural diversity, which ranged from 1.32 to 1.62, exhibited a similar negative influence on PBAI associated with increasing stand density. With regard to pine composition, pine stands with a large proportion of pine basal area, showed higher productivity due to the simple stand structure resulting in better growth of shade intolerant pine. This study found that stand density increases with the development of pine stands and that density increases had negative influences on stand productivity. Collectively, our results suggest that stand density management is essential for increasing stand productivity and carbon sequestration in the Korean red pine-dominated stands of South Korea. • Random forest regression analysis was used to examine factors influencing productivity. • Tree size (QMD) was the most influential factor of pine stand productivity. • QMD, basal area, and structural diversity increased with increasing stand age. • Increases in stand density with stand development decreased stand productivity. • Stand density management is essential to increase productivity of Korean pine stands. [ABSTRACT FROM AUTHOR]

Market capitalization of firms provides valuable information for analyzing stock markets and the size factor is widely used in factor-based investing. Some markets, such as the Korean market, are especially interesting in this respect because they contain extremely large public firms. This study analyzes the effect of the largest firm on factor investing through machine learning models that are effective for variable selection. We demonstrate how machine learning can be used for identifying important factors. Our comparison between US and Korean markets shows the significance of separating the largest firm in analyzing how factors impact performance in the Korean market. • Analyze the Korean stock market, which contains one exceptionally large firm • Compare the effects of the largest public firm between the US and Korea • Use machine learning methods to perform variable selection that includes the effects of the largest firm • Lasso regression and random forest provide feature selection and variable importance for analyzing of market factors • Empirical results reveal unique characteristics of the Korean stock market [ABSTRACT FROM AUTHOR]

This study developed a model to predict concentrations of chlorophyll -a ([Chl-a]) as a proxy for algal population with data from multiple monitoring stations in the Han river basin, by using machine-learning predictive models, then analyzed the relationship between [Chl-a] and the input variables of the optimized model. Daily water quality and meteorological data from 2012 to 2020 were collected from the real-time water quality information system and the meteorological administration of Korea. To quantify model accuracy, the coefficient of determination, root mean square error, and mean absolute error were applied. Among random forest (RF), support vector machine, and artificial neural network, the RF with random dataset showed the highest accuracy. The RF was optimized when 78 trees were applied to the model. Input variables for the best RF model were total organic carbon (feature importance: 27%), total nitrogen (19%), pH (13%), water temperature (8%), total phosphorus (8%), electrical conductivity (7%), dissolved oxygen (6%), minimum air temperature (AT) (4%), mean AT (3%), and maximum AT (3%). The feature-importance analysis showed that total organic carbon was the most important variable to predict [Chl-a] in the Han river basin. Total nitrogen was a more important variable than total phosphorus. [Display omitted] • The optimal input variables were selected through Pearson's correlation analysis. • Random forest (RF) was most suitable for predicting chlorophyll -a ([Chl-a]). • RF had 10 input variables including total organic carbon, nitrogen, and phosphorus. • Total nitrogen was more important than total phosphorus in predicting [Chl-a]. • RF can manage algae associated with multiple monitoring stations in river basin. [ABSTRACT FROM AUTHOR]

• Random forest model predicted largemouth bass occurrence in Han River basin. • Increased largemouth bass occurrence after the 2060 s for the RCP 8.5 scenario. • Expected gradual southeastward shift of largemouth bass to higher elevations. Predicting the shift of invasive species distribution in response to climate change is essential for ecological risk assessment. In this study, the distribution of invasive largemouth bass (Micropterus salmoides) was predicted from 2016 to 2085 in the Han River basin of South Korea using HadGEM3-RA based climate change scenarios (representative concentration pathway [RCP] 4.5 and 8.5 scenarios). A random forest model was developed using largemouth bass occurrence data and environmental variables (temperature, precipitation, flow rate, water quality, elevation, and slope) for five years (2011–2015). The annual mean temperature was the second most contributing variable after elevation. The occurrence of largemouth bass in the Han River basin was expected to increase in the future with increasing annual mean temperature, especially after the 2060 s (2056–2065) in the RCP 8.5 scenario. The latitudinal, longitudinal, and altitudinal shifts of largemouth bass distribution were expected to be −1.76 km, 4.19 km, and 9.07 m per 10 years under the RCP 8.5 scenario, respectively, indicating a gradual southeastward shift to higher locations. Given that invasive fish can disturb the habitat of endemic species, three-dimensional distribution shift under climate change should be considered for the ecological risk assessment of invasive species. [ABSTRACT FROM AUTHOR]

Background: Insulin resistance is a common etiology of metabolic syndrome, but receiver operating characteristic (ROC) curve analysis shows a weak association in Koreans. Using a machine learning (ML) approach, we aimed to generate the best model for predicting insulin resistance in Korean adults aged > 40 of the Ansan/Ansung cohort using a machine learning (ML) approach. Methods: The demographic, anthropometric, biochemical, genetic, nutrient, and lifestyle variables of 8842 participants were included. The polygenetic risk scores (PRS) generated by a genome-wide association study were added to represent the genetic impact of insulin resistance. They were divided randomly into the training (n = 7037) and test (n = 1769) sets. Potentially important features were selected in the highest area under the curve (AUC) of the ROC curve from 99 features using seven different ML algorithms. The AUC target was ≥0.85 for the best prediction of insulin resistance with the lowest number of features. Results: The cutoff of insulin resistance defined with HOMA-IR was 2.31 using logistic regression before conducting ML. XGBoost and logistic regression algorithms generated the highest AUC (0.86) of the prediction models using 99 features, while the random forest algorithm generated a model with 0.82 AUC. These models showed high accuracy and k-fold values (>0.85). The prediction model containing 15 features had the highest AUC of the ROC curve in XGBoost and random forest algorithms. PRS was one of 15 features. The final prediction models for insulin resistance were generated with the same nine features in the XGBoost (AUC = 0.86), random forest (AUC = 0.84), and artificial neural network (AUC = 0.86) algorithms. The model included the fasting serum glucose, ALT, total bilirubin, HDL concentrations, waist circumference, body fat, pulse, season to enroll in the study, and gender. Conclusion: The liver function, regular pulse checking, and seasonal variation in addition to metabolic syndrome components should be considered to predict insulin resistance in Koreans aged over 40 years. [ABSTRACT FROM AUTHOR]

• Some readily available spatial covariates were used to map landslide susceptibility. • The capability of the GMDH model was explored for landslide susceptibility mapping. • Five hybridized models were developed based on metaheuristic optimization approaches. • Only some hybridized models were better than random forest and boosted regression trees. • The hybridized model of particle swarm optimization algorithm had the highest accuracy. Landslide susceptibility (LS) mapping is an essential tool for landslide risk assessment. This study aimed to provide a new approach with better performance for landslide mapping and adopting readily available variables. In addition, it investigates the capability of a state-of-the-art model developed using the group method of data handling (GMDH) to spatially model LS. Furthermore, hybridized models of GMDH were developed using different metaheuristic algorithms. The study area was the Bonghwa region of South Korea, for which an accurate landslide inventory dataset is available. We considered a total of 13 spatial covariates (altitude, slope, aspect, topographic wetness index, valley depth, plan curvature, profile curvature, distance from fault, distance from river, distance from road, land use, density of forest, and lithology were chosen as independent variables). Two benchmark models—random forest and boosted regression trees—were used to compare their results with the standalone GMDH and hybridized models. We compared model accuracy using the two most robust evaluation metrics, root mean square error (RMSE) and area under the receiver operating characteristic curve (AUROC). The validation results showed that hybridized models outperformed the standalone GMDH model. Moreover, the hybridized GMDH-PSO (AUC = 0.83, RMSE = 0.108), GMDH-IWO (AUC = 0.81, RMSE = 0.111), GMDH-BBO (AUC = 0.8; RMSE = 0.12), and GMDH-ICA (AUC = 0.8; RMSE = 0.117) had a better predictive performance than both RF and BRT. Therefore, the proposed approach could successfully produce landslide susceptibility maps using relatively few readily available variables and can be repeated in data-scarce regions. [ABSTRACT FROM AUTHOR]

• Harmonization and quality control of originally heterogenous observational databases. • Large-scale application of machine learning methods in a distributed data network. • Transparent use of open-source software tools and publicly shared analytical code. As a response to the ongoing COVID-19 pandemic, several prediction models in the existing literature were rapidly developed, with the aim of providing evidence-based guidance. However, none of these COVID-19 prediction models have been found to be reliable. Models are commonly assessed to have a risk of bias, often due to insufficient reporting, use of non-representative data, and lack of large-scale external validation. In this paper, we present the Observational Health Data Sciences and Informatics (OHDSI) analytics pipeline for patient-level prediction modeling as a standardized approach for rapid yet reliable development and validation of prediction models. We demonstrate how our analytics pipeline and open-source software tools can be used to answer important prediction questions while limiting potential causes of bias (e.g. , by validating phenotypes, specifying the target population, performing large-scale external validation, and publicly providing all analytical source code). We show step-by-step how to implement the analytics pipeline for the question: 'In patients hospitalized with COVID-19, what is the risk of death 0 to 30 days after hospitalization?'. We develop models using six different machine learning methods in a USA claims database containing over 20,000 COVID-19 hospitalizations and externally validate the models using data containing over 45,000 COVID-19 hospitalizations from South Korea, Spain, and the USA. Our open-source software tools enabled us to efficiently go end-to-end from problem design to reliable Model Development and evaluation. When predicting death in patients hospitalized with COVID-19, AdaBoost, random forest, gradient boosting machine, and decision tree yielded similar or lower internal and external validation discrimination performance compared to L1-regularized logistic regression, whereas the MLP neural network consistently resulted in lower discrimination. L1-regularized logistic regression models were well calibrated. Our results show that following the OHDSI analytics pipeline for patient-level prediction modelling can enable the rapid development towards reliable prediction models. The OHDSI software tools and pipeline are open source and available to researchers from all around the world. [ABSTRACT FROM AUTHOR]

Precipitation is a crucial component of the water cycle and plays a key role in hydrological processes. Recently, satellite-based precipitation products (SPPs) have provided grid-based precipitation with spatiotemporal variability. However, SPPs contain a lot of uncertainty in estimated precipitation, and the spatial resolution of these products is still relatively coarse. To overcome these limitations, this study aims to generate new grid-based daily precipitation based on a combination of rainfall observation data with multiple SPPs for the period of 2003–2017 across South Korea. A Random Forest (RF) machine-learning algorithm model was applied for producing a new merged precipitation product. In addition, several statistical linear merging methods have been adopted to compare with the results achieved from the RF model. To investigate the efficiency of RF, rainfall data from 64 observed Automated Synoptic Observation System (ASOS) installations were collected to analyze the accuracy of products through several continuous as well as categorical indicators. The new precipitation values produced by the merging procedure generally not only report higher accuracy than a single satellite rainfall product but also indicate that RF is more effective than the statistical merging method. Thus, the achievements from this study point out that the RF model might be applied for merging multiple satellite precipitation products, especially in sparse region areas. [ABSTRACT FROM AUTHOR]

Species distribution models (SDMs), in which species occurrences are related to a suite of environmental variables, have been used as a decision-making tool in ecosystem management. Complex machine learning (ML) algorithms that lack interpretability may hinder the use of SDMs for ecological explanations, possibly limiting the role of SDMs as a decision-support tool. To meet the growing demand of explainable MLs, several interpretable ML methods have recently been proposed. Among these methods, SHaply Additive exPlanation (SHAP) has drawn attention for its robust theoretical justification and analytical gains. In this study, the utility of SHAP was demonstrated by the application of SDMs of four benthic macroinvertebrate species. In addition to species responses, the dataset contained 22 environmental variables monitored at 436 sites across five major rivers of South Korea. A range of ML algorithms was employed for model development. Each ML model was trained and optimized using 10-fold cross-validation. Model evaluation based on the test dataset indicated strong model performance, with an accuracy of ≥0.7 in all evaluation metrics for all MLs and species. However, only the random forest algorithm showed a behavior consistent with the known ecology of the investigated species. SHAP presents an integrated framework in which local interpretations that incorporate local interaction effects are combined to represent the global model structure. Consequently, this framework offered a novel opportunity to assess the importance of variables in predicting species occurrence, not only across sites, but also for individual sites. Furthermore, removing interaction effects from variable importance values (SHAP values) clearly revealed non-linear species responses to variations in environmental variables, indicating the existence of ecological thresholds. This study provides guidelines for the use of a new interpretable method supporting ecosystem management. [Display omitted] • SDMs have been used as decision-making tools in ecosystem management. • SHAP was applied to SDMs of macroinvertebrate species in rivers. • Across sites, velocity had the largest positive effect on species occurrence. • Magnitude and direction of a variable's effect were assessed for individual sites. • SHAP can be used to evaluate models and provide management support. [ABSTRACT FROM AUTHOR]

• A data-driven approach to identify a list of risk factors affecting severities of freeway tunnel crashes. • Combining random forest algorithm with standard binomial regression. • Quantitatively confirming national tunnel traffic safety improvement strategies. • Recommending additional strategy specifications for freeway tunnel safety enhancement. Freeway tunnel strikes by motor vehicles inflict serious damages to the infrastructure, cause personal injuries, and create traffic congestions. Freeway tunnel hits are a constant threat in South Korea due to its mostly mountainous terrain. Despite efforts by public agencies to include disaster remedial or preventative facilities in the development of freeway tunnels, these facilities are designed mainly to reduce the number of collisions instead of also mitigate the consequences of a crash. Hence, the Korea Ministry of Land, Infrastructure and Transport (KMOLIT) recently presented a plan to modify the list of risk factors affecting tunnel traffic safety, and it recommended several strategies for tunnel traffic safety management. The study presented here took a data-driven approach to quantitatively confirming the strategies recommended by KMOLIT through the random forest-based binomial regression. The following factors were found to be significantly associated with serious injury crashes involving freeway tunnel strikes: adverse weather, fatigued and distracted drivers, collision type (i.e., head-on/angle/rear-end), tunnel exit, tunnel width, curve radius (radius less than 1800 m), and heavy vehicle. This study compares specifications of each government strategy with the effects of the identified risk factors on injuries involved in tunnel crashes to quantitatively support recommendations to modify the government strategies. [ABSTRACT FROM AUTHOR]

The demand for rice production in Asia is expected to increase by 70% in the next 30 years, which makes evident the need for a balanced productivity and effective food security management at a national and continental level. Consequently, the timely and accurate mapping of paddy rice extent and its productivity assessment is of utmost significance. In turn, this requires continuous area monitoring and large scale mapping, at the parcel level, through the processing of big satellite data of high spatial resolution. This work designs and implements a paddy rice mapping pipeline in South Korea that is based on a time-series of Sentinel-1 and Sentinel-2 data for the year of 2018. There are two challenges that we address; the first one is the ability of our model to manage big satellite data and scale for a nationwide application. The second one is the algorithm's capacity to cope with scarce labeled data to train supervised machine learning algorithms. Specifically, we implement an approach that combines unsupervised and supervised learning. First, we generate pseudo-labels for rice classification from a single site (Seosan-Dangjin) by using a dynamic k-means clustering approach. The pseudo-labels are then used to train a Random Forest (RF) classifier that is fine-tuned to generalize in two other sites (Haenam and Cheorwon). The optimized model was then tested against 40 labeled plots, evenly distributed across the country. The paddy rice mapping pipeline is scalable as it has been deployed in a High Performance Data Analytics (HPDA) environment using distributed implementations for both k-means and RF classifiers. When tested across the country, our model provided an overall accuracy of 96.69% and a kappa coefficient 0.87. Even more, the accurate paddy rice area mapping was returned early in the year (late July), which is key for timely decision-making. Finally, the performance of the generalized paddy rice classification model, when applied in the sites of Haenam and Cheorwon, was compared to the performance of two equivalent models that were trained with locally sampled labels. The results were comparable and highlighted the success of the model's generalization and its applicability to other regions. [ABSTRACT FROM AUTHOR]

In this study, the visibility of South Korea was predicted (VISRF) using a random forest (RF) model based on ground observation data from the Automated Synoptic Observing System (ASOS) and air pollutant data from the European Centre for Medium-Range Weather Forecasts (ECMWF) Copernicus Atmosphere Monitoring Service (CAMS) model. Visibility was predicted and evaluated using a training set for the period 2017–2018 and a test set for 2019. VISRF results were compared and analyzed using visibility data from the ASOS (VISASOS) and the Unified Model (UM) Local Data Assimilation and Prediction System (LDAPS) (VISLDAPS) operated by the Korea Meteorological Administration (KMA). Bias, root mean square error (RMSE), and correlation coefficients (R) for the VISASOS and VISLDAPS datasets were 3.67 km, 6.12 km, and 0.36, respectively, compared to 0.14 km, 2.84 km, and 0.81, respectively, for the VISASOS and VISRF datasets. Based on these comparisons, the applied RF model offers significantly better predictive performance and more accurate visibility data (VISRF) than the currently available VISLDAPS outputs. This modeling approach can be implemented by authorities to accurately estimate visibility and thereby reduce accidents, risks to public health, and economic losses, as well as inform on urban development policies and environmental regulations. [ABSTRACT FROM AUTHOR]

It is necessary to monitor, acquire, preprocess, and classify microseismic data to understand active faults or other causes of earthquakes, thereby facilitating the preparation of early-warning earthquake systems. Accordingly, this study proposes the application of machine learning for signal–noise classification of microseismic data from Pohang, South Korea. For the first time, unique microseismic data were obtained from the monitoring system of the borehole station PHBS8 located in Yongcheon-ri, Pohang region, while hydraulic stimulation was being conducted. The collected data were properly preprocessed and utilized as training and test data for supervised and unsupervised learning methods: random forest, convolutional neural network, and K-medoids clustering with fast Fourier transform. The supervised learning methods showed 100% and 97.4% of accuracy for the training and test data, respectively. The unsupervised method showed 97.0% accuracy. Consequently, the results from machine learning validated that automation based on the proposed supervised and unsupervised learning applications can classify the acquired microseismic data in real time. [ABSTRACT FROM AUTHOR]

• We used a random forest algorithm to create a predictive model for depression. • The performance was good with an AUC of 0.839, sensitivity of 0.723, and specificity of 0.824. • Prescription of benzodiazepines had the greatest influence on the future onset of depression. Depression is a mental illness that causes significant disturbances in daily life. Depression is commonly associated with low mood, severe health problems, and substantial socioeconomic burden; hence, it is necessary to be able to detect depression earlier. We utilized the medical check-up cohort database of the National Health Insurance Sharing Service in Korea. We split the total dataset into training (70%) and test (30%) sets. Subsequently, five-fold cross validation was performed in the training set. The holdout test set was only used in the last step to evaluate the performance of the predictive model. Random forest algorithm was used for the predictive model. The analysis included 433,190 individuals who had a national medical check-up from 2009–2015, which included 10,824 (2.56%) patients in the depression group. The area under the receiver-operating curve was 0.849. Other performance metrics included a sensitivity of 0.737, specificity of 0.824, positive predictive value of 0.097, negative predictive value of 0.992, and accuracy of 0.780. Our predictive model could contribute to proactively reducing depression prevalence by administering interventions to prevent depression in patients receiving medical check-up. Future studies are needed to prospectively validate the predictability of this model. [ABSTRACT FROM AUTHOR]

Landslides impact on human activities and socio-economic development, especially in mountainous areas. This study focuses on the comparison of the prediction capability of advanced machine learning techniques for the rainfall-induced shallow landslide susceptibility of Deokjeokri catchment and Karisanri catchment in South Korea. The influencing factors for landslides, i.e., topographic, hydrologic, soil, forest, and geologic factors, are prepared from various sources based on availability, and a multicollinearity test is also performed to select relevant causative factors. The landslide inventory maps of both catchments are obtained from historical information, aerial photographs and performed field surveys. In this study, Deokjeokri catchment is considered as a training area and Karisanri catchment as a testing area. The landslide inventories contain 748 landslide points in training and 219 points in testing areas. Three landslide susceptibility maps using machine learning models, i.e., Random Forest (RF), Extreme Gradient Boosting (XGBoost) and Deep Neural Network (DNN), are prepared and compared. The outcomes of the analyses are validated using the landslide inventory data. A receiver operating characteristic curve (ROC) method is used to verify the results of the models. The results of this study show that the training accuracy of RF is 0.756 and the testing accuracy is 0.703. Similarly, the training accuracy of XGBoost is 0.757 and testing accuracy is 0.74. The prediction of DNN revealed acceptable agreement between the susceptibility map and the existing landslides, with a training accuracy of 0.855 and testing accuracy of 0.802. The results showed that the DNN model achieved lower prediction error and higher accuracy results than other models for shallow landslide modeling in the study area. [ABSTRACT FROM AUTHOR]