• The Transactions of the Korean Institute of Power Electronics
• /
• v.23 no.6
• /
• pp.403-407
• /
• 2018

This study proposes a predictive direct power control method in a modular multilevel converter (MMC) high-voltage direct-current (HVDC) system. The conventional proportional integral (PI)-based control method uses a cascaded connection and requires an optimal gain selection procedure and additional decoupling scheme. However, the proposed control method has a simple structure for active/reactive power control due to the direct power control scheme and exhibits a fast dynamic response by predicting the future status of system variables and considering time delay. The effectiveness of the proposed method is verified by simulation results.

• Journal of Information Processing Systems
• /
• v.17 no.3
• /
• pp.630-644
• /
• 2021

Educational data analysis is attracting increasing attention with the rise of the big data industry. The amounts and types of learning data available are increasing steadily, and the information technology required to analyze these data continues to develop. The early identification of potential dropout students is very important; education is important in terms of social movement and social achievement. Here, we analyze educational data and generate predictive models for student dropout using logistic regression, a decision tree, a naïve Bayes method, and a multilayer perceptron. The multilayer perceptron model using independent variables selected via the variance analysis showed better performance than the other models. In addition, we experimentally found that not only grades but also extracurricular activities were important in terms of preventing student dropout.

• Journal of Institute of Control, Robotics and Systems
• /
• v.3 no.3
• /
• pp.272-279
• /
• 1997

In this paper, a fuzzy generalized predictive control (FGPC) for non-linear plants is proposed. In the proposed method, the receding horizon control is applied to the control part, while fuzzy systems are used for the predictor part. It is suggested that the fuzzy predictor is time-varying affine with respect to input variables for easy computation of control inputs. Since the receding horizon control can be obtained only with a predictor instead of a plant model, the fuzzy predictor is obtained directly from input-output data without identifying a plant model. A parameter estimation algorithm is used for identifying the fuzzy predictor. The control inputs of the FGPC are computed by minimizing a receding horizon cost function with predicted plant outputs. The proposed controller has a similar architecture to the generalized predictive control (GPC) except for the predictor synthesis method, and thus may possess inherent good properties of the GPC. Computer simulations show that the performance of the FGPC is satisfactory.

• Journal of Korean Society for Quality Management
• /
• v.48 no.3
• /
• pp.511-520
• /
• 2020

Purpose: The purpose of this study was to devise an accurate machine learning model for predicting silica concentrations following the addition of impurities, through time series analysis of mining data. Methods: The mining data were preprocessed and subjected to time series analysis using the machine learning model. Through correlation analysis, valid variables were selected and meaningless variables were excluded. To reflect changes over time, dependent variables at baseline were treated as independent variables at later time points. The relationship between independent variables and the dependent variable after n point was subjected to Pearson correlation analysis. Results: The correlation (R2) was strongest after 3 hours, which was adopted as a dependent variable. According to root mean square error (RMSE) data, the proposed method was superior to the other machine learning methods. The XGboost algorithm showed the best predictive performance. Conclusion: This study is important given the current lack of machine learning studies pertaining to the domestic mining industry. In addition, using time series analysis in mining data will show further improvement. Before establishing a predictive model for the proposed method, predictions should be made using data with time series characteristics. After doing this work, it should also improve prediction accuracy in other domains.

• Journal of Internet Computing and Services
• /
• v.23 no.3
• /
• pp.125-135
• /
• 2022

This study aims to explore variables using machine learning and provide analysis techniques suitable for predicting pharmacy sales whether government statistical indicators built to create an industrial ecosystem based on data, network, and artificial intelligence affect pharmacy sales. Therefore, this study explored predictive variables and performance through machine learning techniques such as Random Forest, XGBoost, LightGBM, and CatBoost using analysis data from January 2016 to December 2021 for 28 government statistical indicators and pharmacies in the retail sector. As a result of the analysis, economic sentiment index, economic accompanying index circulation change, and consumer sentiment index, which are economic indicators, were found to be important variables affecting pharmacy sales. As a result of examining the indicators MAE, MSE, and RMSE for regression performance, random forests showed the best performance than XGBoost, LightGBM, and CatBoost. Therefore, this study presented variables and optimal machine learning techniques that affect pharmacy sales based on machine learning results, and proposed several implications and follow-up studies.

• Journal of Preventive Medicine and Public Health
• /
• v.34 no.3
• /
• pp.237-243
• /
• 2001

Objectives : To examine the screening rate of cervical cancer in women and to find out the predictive factors for participation in cervical cancer screening programs within their life-time and within the last two years. Methods : The data was based on self-reported questionnaires from 1,613 women whose ages ranged from 26 to 60 years; this survey was peformed between December 1999 and January 2000. This study analyzed the predictive factors for participation in cervical cancer screening programs within their life-time and within the last two years. A logistic regression analysis was performed in order to derive the significant variables from the predisposing factors(demographic factor, health promotion behavior, reproductive factor), intervention factors(information channel, relation with medical stan, and proximal factors(attitude, social influence, self-efficacy). All analyses were peformed by the PC-SAS 6.12. Results : Our analyses showed that the screening rate for the women who received a cervical cancer screening(Pap smear) more than once within their life-time was 56.1% while those who had received one within the last two years was 34.5%. The significant factors for participation in cervical cancer screening program within their life-time were their income, married age, health promotion score, relation with medical staffs, social influence, and self-efficacy. On the other hand, age, number of pregnancies, menarche age, relation with medical staffs, social influences, and self-efficacy were significant factors for those being screened within the last two years. The predictive power of the logit model within their life-time was 68.8% and that within the last two years was 66.6%. Conclusion : The predictive factors for participation in cervical cancer screening program within their life-time are different from those for within the last two years. and that women's relations with medical staffs and social influences were the critical factors impacting on cervical cancer screening rates.

• Journal of the Korea Convergence Society
• /
• v.11 no.6
• /
• pp.203-208
• /
• 2020

This study examined whether Mild Cognitive Impairment (MCI) is related to the reduction of specific memory among linguistic memory and visuospatial memory, and to identify the most predictive index for discriminating MCI from normal elderly. The subjects were analyzed for 189 elderly (103 healthy elderly, 86 MCI). The verbal memory was used by the Seoul Verbal Learning Test. visuospatial memory was measured using the Rey Complex Figure Test. As a result of multiple logistic regression, verbal memory and visuospatial memory showed significant predictive performance in discriminating MCI from normal elderly. On the other hand, when all the confounding variables were corrected, including the results of each memory test, the predictive power was significant in distinguishing MCI from normal aging only in the immediate recall of verbal memory, and the predictive power was not significant in the immediate recall of visuospatial memory. This result suggests that delayed recall of visuospatial memory and immediate recall of verbal memory are the best combinations to discriminate memory ability of MCI.

• The Journal of Korean Medicine
• /
• v.40 no.4
• /
• pp.49-60
• /
• 2019

Objectives: To develop a standardized diagnostic pattern identification equation for stroke patients, our group conducted a study to derive the predictive logistic equations. However, the sample size was relatively small. In the current study, we aimed to derive new predictive logistic equations for each diagnostic pattern using an expanded number of subjects. Methods: This study was a hospital-based multi-center trial recruited stroke patients within 30 days of symptom onset. Patients' general information, and the variables related to diagnostic pattern identification were measured. The diagnostic pattern of each patient was identified independently by two Korean Medicine Doctors. To derive a predictive model for pattern identification, binary logistic regression analysis was applied. Results: Among the 1,251 patients, 385 patients (30.8%) had the Fire Heat Pattern, 460 patients (36.8%) the Phlegm Dampness Pattern, 212 patients (16.9%) the Qi Deficiency Pattern, and 194 patients (15.5%) the Yin Deficiency Pattern. After the regression analysis, the predictive logistic equations for each pattern were determined. Conclusion: The predictive equations for Fire Heat, Phlegm Dampness, Qi Deficiency, and Yin Deficiency would be useful to determine individual stroke patients' pattern identification in the clinical setting. However, further studies using objective measurements are necessary to validate these data.

• Korean Journal of Community Nutrition
• /
• v.19 no.6
• /
• pp.568-580
• /
• 2014

Objectives: The purpose of this study was to measure the resting metabolic rate (RMR) and to assess the accuracy of RMR predictive equations for Korean farmers. Methods: Subjects were 161 healthy Korean farmers (50 males, 111 females) in Gangwon-area. The RMR was measured by indirect calorimetry for 20 minutes following a 12-hour overnight fasting. Selected predictive equations were Harris-Benedict, Mifflin, Liu, KDRI, Cunningham (1980, 1991), Owen-W, F, FAO/WHO/UNU-W, WH, Schofield-W, WH, Henry-W, WH. The accuracy of the equations was evaluated on the basis of bias, RMSPE, accurate prediction and Bland-Altman plot. Further, new RMR predictive equations for the subjects were developed by multiple regression analysis using the variables highly related to RMR. Results: The mean of the measured RMR was 1703 kcal/day in males and 1343 kcal/day in females. The Cunningham (1980) equation was the closest to measured RMR than others in males and in females (males Bias -0.47%, RMSPE 110 kcal/day, accurate prediction 80%, females Bias 1.4%, RMSPE 63 kcal/day, accurate prediction 81%). Body weight, BMI, circumferences of waist and hip, fat mass and FFM were significantly correlated with measured RMR. Thus, derived prediction equation as follow : males RMR = 447.5 + 17.4 Wt, females RMR = 684.5 - 3.5 Ht + 11.8 Wt + 12.4 FFM. Conclusions: This study showed that Cunningham (1980) equation was the most accurate to predict RMR of the subjects. Thus, Cunningham (1980) equation could be used to predict RMR of Korean farmers studied in this study. Future studies including larger subjects should be carried out to develop RMR predictive equations for Korean farmers.

• Communications for Statistical Applications and Methods
• /
• v.30 no.6
• /
• pp.577-587
• /
• 2023

Conventional categorical data imputation techniques, such as mode imputation, often encounter issues related to overestimation. If the variable has too many categories, multinomial logistic regression imputation method may be impossible due to computational limitations. To rectify these limitations, we propose a two-stage imputation method. During the first stage, we utilize the Boruta variable selection method on the complete dataset to identify significant variables for the target categorical variable. Then, in the second stage, we use the important variables for the target categorical variable for logistic regression to impute missing data in binary variables, polytomous regression to impute missing data in categorical variables, and predictive mean matching to impute missing data in quantitative variables. Through analysis of both asymmetric and non-normal simulated and real data, we demonstrate that the two-stage imputation method outperforms imputation methods lacking variable selection, as evidenced by accuracy measures. During the analysis of real survey data, we also demonstrate that our suggested two-stage imputation method surpasses the current imputation approach in terms of accuracy.