• Nuclear Engineering and Technology
• /
• v.53 no.12
• /
• pp.4022-4032
• /
• 2021

In this study, machine learning (ML) techniques were used to model surveillance test data of nuclear power plants from an international database of the ASTM E10.02 committee. Regression modeling was conducted using various techniques, including Cubist, XGBoost, and a support vector machine. The root mean square deviation of each ML model for the baseline dataset was less than that of the ASTM E900-15 nonlinear regression model. With respect to the interpolation, the ML methods provided excellent predictions with relatively few computations when applied to the given data range. The effect of the explanatory variables on the transition temperature shift (TTS) for the ML methods was analyzed, and the trends were slightly different from those for the ASTM E900-15 model. ML methods showed some weakness in the extrapolation of the fluence in comparison to the ASTM E900-15, while the Cubist method achieved an extrapolation to a certain extent. To achieve a more reliable prediction of the TTS, it was confirmed that advanced techniques should be considered for extrapolation when applying ML modeling.

• Korean Journal of Soil Science and Fertilizer
• /
• v.47 no.5
• /
• pp.319-323
• /
• 2014

This study investigates the prediction of soil chemical properties (organic matter (OM), pH, Ca, Mg, K, Na, total acidity, cation exchange capacity (CEC)) on 688 Korean soil samples using the visible-near infrared reflectance (VIS-NIR) spectroscopy. Reflectance from the visible to near-infrared spectrum (350 to 2500 nm) was acquired using the ASD Field Spec Pro. A total of 688 soil samples from 168 soil profiles were collected from 2009 to 2011. The spectra were resampled to 10 nm spacing and converted to the 1st derivative of absorbance (log (1/R)), which was used for predicting soil chemical properties. Principal components analysis (PCA), partial least squares regression (PLSR) and regression rules model (Cubist) were applied to predict soil chemical properties. The regression rules model (Cubist) showed the best results among these, with lower error on the calibration data. For quantitatively determining OM, total acidity, CEC, a VIS-NIR spectroscopy could be used as a routine method if the estimation quality is more improved.

• Journal of Digital Convergence
• /
• v.19 no.7
• /
• pp.271-282
• /
• 2021

This study was conducted to develop a model for predicting the length of stay for premature infants through machine learning. For the development of this model, 6,149 cases of premature infants discharged from the hospital from 2011 to 2016 of the discharge injury in-depth survey data collected by the Korea Centers for Disease Control and Prevention were used. The neural network model of the initial hospitalization was superior to other models with an explanatory power (R²) of 0.75. In the model added by converting the clinical diagnosis to CCS(Clinical class ification software), the explanatory power (R²) of the cubist model was 0.81, which was superior to the random forest, gradient boost, neural network, and penalty regression models. In this study, using national data, a model for predicting the length of stay for premature infants was presented through machine learning and its applicability was confirmed. However, due to the lack of clinical information and parental information, additional research is needed to improve future performance.

• Korean Journal of Agricultural Science
• /
• v.47 no.4
• /
• pp.1097-1107
• /
• 2020

Assessment of soil carbon stock is essential for climate change mitigation and soil fertility. The digital soil mapping (DSM) is well known as a general technique to estimate the soil carbon stocks and upgrade previous soil maps. The aim of this study is to calculate the soil carbon stock in the top soil layer (0 to 30 cm) in Jeolla Province of South Korea using the DSM technique. To predict spatial carbon stock, we used Cubist, which a data-mining algorithm model base on tree regression. Soil samples (130 in total) were collected from three depths (0 to 10 cm, 10 to 20 cm, 20 to 30 cm) considering spatial distribution in Jeolla Province. These data were randomly divided into two sets for model calibration (70%) and validation (30%). The results showed that clay content, topographic wetness index (TWI), and digital elevation model (DEM) were the most important environmental covariate predictors of soil carbon stock. The predicted average soil carbon density was 3.88 kg·m-2. The R2 value representing the model's performance was 0.6, which was relatively high compared to a previous study. The total soil carbon stocks at a depth of 0 to 30 cm in Jeolla Province were estimated to be about 81 megatons.

• Korean Journal of Soil Science and Fertilizer
• /
• v.45 no.4
• /
• pp.497-502
• /
• 2012

This study investigates the prediction of soil OM on Korean soils using the Visible-Near Infrared (Vis-NIR) spectroscopy. The ASD Field Spec Pro was used to acquire the reflectance of soil samples to visible to near-infrared radiation (350 to 2500 nm). A total of 503 soil samples from 61 Korean soil series were scanned using the instrument and OM was measured using the Walkley and Black method. For data analysis, the spectra were resampled from 500-2450 nm with 4 nm spacing and converted to the $1^{st}$ derivative of absorbance (log (1/R)). Partial least squares regression (PLSR) and regression rules model (Cubist) were applied to predict soil OM. Regression rules model estimates the target value by building conditional rules, and each rule contains a linear expression predicting OM from selected absorbance values. The regression rules model was shown to give a better prediction compared to PLSR. Although the prediction for Andisols had a larger error, soil order was not found to be useful in stratifying the prediction model. The stratification used by Cubist was mainly based on absorbance at wavelengths of 850 and 2320 nm, which corresponds to the organic absorption bands. These results showed that there could be more information on soil properties useful to classify or group OM data from Korean soils. In conclusion, this study shows it is possible to develop good prediction model of OM from Korean soils and provide data to reexamine the existing prediction models for more accurate prediction.