• The Korean Journal of Applied Statistics
• /
• v.31 no.3
• /
• pp.367-382
• /
• 2018

Prediction models for a corporate bond rating in existing studies have been developed using various models such as linear regression, ordered logit, and random forest. Financial characteristics help build prediction models that are expected to be contained in the assigning model of the bond rating agencies. However, the ranges of bond ratings in existing studies vary from 5 to 20 and the prediction models were developed with samples in which the target companies and the observation periods are different. Thus, a simple comparison of the prediction accuracies in each study cannot determine the best prediction model. In order to conduct a fair comparison, this study has collected corporate bond ratings and financial characteristics from 2013 to 2017 and applied prediction models to them. In addition, we applied the elastic-net penalty for the linear regression, the ordered logit, and the ordered probit. Our comparison shows that data-driven variable selection using the elastic-net improves prediction accuracy in each corresponding model, and that the random forest is the most appropriate model in terms of prediction accuracy, which obtains 69.6% accuracy of the exact rating prediction on average from the 5-fold cross validation.

• Safety and Health at Work
• /
• v.10 no.4
• /
• pp.452-460
• /
• 2019

Background: Monitoring and control of PM_2.5 are being recognized as key to address health issues attributed to PM_2.5. Availability of low-cost PM_2.5 sensors made it possible to introduce a number of portable PM_2.5 monitors based on light scattering to the consumer market at an affordable price. Accuracy of light scatteringe-based PM_2.5 monitors significantly depends on the method of calibration. Static calibration curve is used as the most popular calibration method for low-cost PM_2.5 sensors particularly because of ease of application. Drawback in this approach is, however, the lack of accuracy. Methods: This study discussed the calibration of a low-cost PM_2.5-monitoring device (PMD) to improve the accuracy and reliability for practical use. The proposed method is based on construction of the PM_2.5 sensor network using Message Queuing Telemetry Transport (MQTT) protocol and web query of reference measurement data available at government-authorized PM monitoring station (GAMS) in the republic of Korea. Four machine learning (ML) algorithms such as support vector machine, k-nearest neighbors, random forest, and extreme gradient boosting were used as regression models to calibrate the PMD measurements of PM_2.5. Performance of each ML algorithm was evaluated using stratified K-fold cross-validation, and a linear regression model was used as a reference. Results: Based on the performance of ML algorithms used, regression of the output of the PMD to PM_2.5 concentrations data available from the GAMS through web query was effective. The extreme gradient boosting algorithm showed the best performance with a mean coefficient of determination (R²) of 0.78 and standard error of 5.0 ㎍/㎥, corresponding to 8% increase in R² and 12% decrease in root mean square error in comparison with the linear regression model. Minimum 100 hours of calibration period was found required to calibrate the PMD to its full capacity. Calibration method proposed poses a limitation on the location of the PMD being in the vicinity of the GAMS. As the number of the PMD participating in the sensor network increases, however, calibrated PMDs can be used as reference devices to nearby PMDs that require calibration, forming a calibration chain through MQTT protocol. Conclusions: Calibration of a low-cost PMD, which is based on construction of PM_2.5 sensor network using MQTT protocol and web query of reference measurement data available at a GAMS, significantly improves the accuracy and reliability of a PMD, thereby making practical use of the low-cost PMD possible.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.26 no.2
• /
• pp.93-98
• /
• 2016

For accurate precipitation forecasts the choice of weather factors and prediction method is very important. Recently, machine learning has been widely used for forecasting precipitation, and artificial neural network, one of machine learning techniques, showed good performance. In this paper, we suggest a new method for forecasting precipitation using DBN, one of deep learning techniques. DBN has an advantage that initial weights are set by unsupervised learning, so this compensates for the defects of artificial neural networks. We used past precipitation, temperature, and the parameters of the sun and moon's motion as features for forecasting precipitation. The dataset consists of observation data which had been measured for 40 years from AWS in Seoul. Experiments were based on 8-fold cross validation. As a result of estimation, we got probabilities of test dataset, so threshold was used for the decision of precipitation. CSI and Bias were used for indicating the precision of precipitation. Our experimental results showed that DBN performed better than MLP.

• The Korean Journal of Applied Statistics
• /
• v.20 no.3
• /
• pp.423-440
• /
• 2007

Although various models have been developed to establish the enterprise credit scoring, no model has utilized the enterprise human resource so far. The purpose of this study was to build an enterprise credit scoring model using enterprise human resource factors. The data to measure the enterprise credit score were made by the first-year research material of HCCP was used to investigate the enterprise human resource and 2004 Credit Rating Score generated from KIS-Credit Scoring Model. The independent variables were chosen among questionnaires of HCCP based on Mclagan(1989)'s HR wheel model, and the credit score of Korean Information Service was used for the dependent variables. The statistical method used for data analysis was logistic regression. As a result of constructing a model, 22 variables were selected. To see these specifically by each large area, 6 variables in human resource development(HRD) area, 15 in human resource management(HRM) area, and 1 in the other area were chosen. As a consequence of 10 fold cross validation, misclassification rate and G-mean were 30.81 and 68.27 respectively. Decile having the highest response rate was bigger than the one having the lowest response rate by 6.08 times, and had a tendency to decrease. Therefore, the result of study showed that the proposed model was appropriate to measure enterprise credit score using enterprise human resource variables.

• The Journal of Engineering Geology
• /
• v.31 no.3
• /
• pp.395-432
• /
• 2021

Groundwater recharge rates vary widely by location and with time. They are difficult to measure directly and are thus often estimated using simulations. This study employed frequency and regression analysis and a classification and regression tree (CART) algorithm in a machine learning method to estimate groundwater recharge. CART algorithms are considered for the distribution of precipitation by subbasin (PCP), geomorphological data, indices of the relationship between vegetation and landuse, and soil type. The considered geomorphological data were digital elevaion model (DEM), surface slope (SLOP), surface aspect (ASPT), and indices were the perpendicular vegetation index (PVI), normalized difference vegetation index (NDVI), normalized difference tillage index (NDTI), normalized difference residue index (NDRI). The spatio-temperal distribution of groundwater recharge in the SWAT-MOD-FLOW program, was classified as group 4, run in R, sampled for random and a model trained its groundwater recharge was predicted by CART condidering modified PVI, NDVI, NDTI, NDRI, PCP, and geomorphological data. To assess inter-rater reliability for group 4 groundwater recharge, the Kappa coefficient and overall accuracy and confusion matrix using K-fold cross-validation were calculated. The model obtained a Kappa coefficient of 0.3-0.6 and an overall accuracy of 0.5-0.7, indicating that the proposed model for estimating groundwater recharge with respect to soil type and vegetation cover is quite reliable.

• Journal of Intelligence and Information Systems
• /
• v.26 no.1
• /
• pp.23-45
• /
• 2020

Big data is creating in a wide variety of fields such as medical care, manufacturing, logistics, sales site, SNS, and the dataset characteristics are also diverse. In order to secure the competitiveness of companies, it is necessary to improve decision-making capacity using a classification algorithm. However, most of them do not have sufficient knowledge on what kind of classification algorithm is appropriate for a specific problem area. In other words, determining which classification algorithm is appropriate depending on the characteristics of the dataset was has been a task that required expertise and effort. This is because the relationship between the characteristics of datasets (called meta-features) and the performance of classification algorithms has not been fully understood. Moreover, there has been little research on meta-features reflecting the characteristics of multi-class. Therefore, the purpose of this study is to empirically analyze whether meta-features of multi-class datasets have a significant effect on the performance of classification algorithms. In this study, meta-features of multi-class datasets were identified into two factors, (the data structure and the data complexity,) and seven representative meta-features were selected. Among those, we included the Herfindahl-Hirschman Index (HHI), originally a market concentration measurement index, in the meta-features to replace IR(Imbalanced Ratio). Also, we developed a new index called Reverse ReLU Silhouette Score into the meta-feature set. Among the UCI Machine Learning Repository data, six representative datasets (Balance Scale, PageBlocks, Car Evaluation, User Knowledge-Modeling, Wine Quality(red), Contraceptive Method Choice) were selected. The class of each dataset was classified by using the classification algorithms (KNN, Logistic Regression, Nave Bayes, Random Forest, and SVM) selected in the study. For each dataset, we applied 10-fold cross validation method. 10% to 100% oversampling method is applied for each fold and meta-features of the dataset is measured. The meta-features selected are HHI, Number of Classes, Number of Features, Entropy, Reverse ReLU Silhouette Score, Nonlinearity of Linear Classifier, Hub Score. F1-score was selected as the dependent variable. As a result, the results of this study showed that the six meta-features including Reverse ReLU Silhouette Score and HHI proposed in this study have a significant effect on the classification performance. (1) The meta-features HHI proposed in this study was significant in the classification performance. (2) The number of variables has a significant effect on the classification performance, unlike the number of classes, but it has a positive effect. (3) The number of classes has a negative effect on the performance of classification. (4) Entropy has a significant effect on the performance of classification. (5) The Reverse ReLU Silhouette Score also significantly affects the classification performance at a significant level of 0.01. (6) The nonlinearity of linear classifiers has a significant negative effect on classification performance. In addition, the results of the analysis by the classification algorithms were also consistent. In the regression analysis by classification algorithm, Naïve Bayes algorithm does not have a significant effect on the number of variables unlike other classification algorithms. This study has two theoretical contributions: (1) two new meta-features (HHI, Reverse ReLU Silhouette score) was proved to be significant. (2) The effects of data characteristics on the performance of classification were investigated using meta-features. The practical contribution points (1) can be utilized in the development of classification algorithm recommendation system according to the characteristics of datasets. (2) Many data scientists are often testing by adjusting the parameters of the algorithm to find the optimal algorithm for the situation because the characteristics of the data are different. In this process, excessive waste of resources occurs due to hardware, cost, time, and manpower. This study is expected to be useful for machine learning, data mining researchers, practitioners, and machine learning-based system developers. The composition of this study consists of introduction, related research, research model, experiment, conclusion and discussion.

• Journal of Intelligence and Information Systems
• /
• v.18 no.2
• /
• pp.29-45
• /
• 2012

Bond rating is regarded as an important event for measuring financial risk of companies and for determining the investment returns of investors. As a result, it has been a popular research topic for researchers to predict companies' credit ratings by applying statistical and machine learning techniques. The statistical techniques, including multiple regression, multiple discriminant analysis (MDA), logistic models (LOGIT), and probit analysis, have been traditionally used in bond rating. However, one major drawback is that it should be based on strict assumptions. Such strict assumptions include linearity, normality, independence among predictor variables and pre-existing functional forms relating the criterion variablesand the predictor variables. Those strict assumptions of traditional statistics have limited their application to the real world. Machine learning techniques also used in bond rating prediction models include decision trees (DT), neural networks (NN), and Support Vector Machine (SVM). Especially, SVM is recognized as a new and promising classification and regression analysis method. SVM learns a separating hyperplane that can maximize the margin between two categories. SVM is simple enough to be analyzed mathematical, and leads to high performance in practical applications. SVM implements the structuralrisk minimization principle and searches to minimize an upper bound of the generalization error. In addition, the solution of SVM may be a global optimum and thus, overfitting is unlikely to occur with SVM. In addition, SVM does not require too many data sample for training since it builds prediction models by only using some representative sample near the boundaries called support vectors. A number of experimental researches have indicated that SVM has been successfully applied in a variety of pattern recognition fields. However, there are three major drawbacks that can be potential causes for degrading SVM's performance. First, SVM is originally proposed for solving binary-class classification problems. Methods for combining SVMs for multi-class classification such as One-Against-One, One-Against-All have been proposed, but they do not improve the performance in multi-class classification problem as much as SVM for binary-class classification. Second, approximation algorithms (e.g. decomposition methods, sequential minimal optimization algorithm) could be used for effective multi-class computation to reduce computation time, but it could deteriorate classification performance. Third, the difficulty in multi-class prediction problems is in data imbalance problem that can occur when the number of instances in one class greatly outnumbers the number of instances in the other class. Such data sets often cause a default classifier to be built due to skewed boundary and thus the reduction in the classification accuracy of such a classifier. SVM ensemble learning is one of machine learning methods to cope with the above drawbacks. Ensemble learning is a method for improving the performance of classification and prediction algorithms. AdaBoost is one of the widely used ensemble learning techniques. It constructs a composite classifier by sequentially training classifiers while increasing weight on the misclassified observations through iterations. The observations that are incorrectly predicted by previous classifiers are chosen more often than examples that are correctly predicted. Thus Boosting attempts to produce new classifiers that are better able to predict examples for which the current ensemble's performance is poor. In this way, it can reinforce the training of the misclassified observations of the minority class. This paper proposes a multiclass Geometric Mean-based Boosting (MGM-Boost) to resolve multiclass prediction problem. Since MGM-Boost introduces the notion of geometric mean into AdaBoost, it can perform learning process considering the geometric mean-based accuracy and errors of multiclass. This study applies MGM-Boost to the real-world bond rating case for Korean companies to examine the feasibility of MGM-Boost. 10-fold cross validations for threetimes with different random seeds are performed in order to ensure that the comparison among three different classifiers does not happen by chance. For each of 10-fold cross validation, the entire data set is first partitioned into tenequal-sized sets, and then each set is in turn used as the test set while the classifier trains on the other nine sets. That is, cross-validated folds have been tested independently of each algorithm. Through these steps, we have obtained the results for classifiers on each of the 30 experiments. In the comparison of arithmetic mean-based prediction accuracy between individual classifiers, MGM-Boost (52.95%) shows higher prediction accuracy than both AdaBoost (51.69%) and SVM (49.47%). MGM-Boost (28.12%) also shows the higher prediction accuracy than AdaBoost (24.65%) and SVM (15.42%)in terms of geometric mean-based prediction accuracy. T-test is used to examine whether the performance of each classifiers for 30 folds is significantly different. The results indicate that performance of MGM-Boost is significantly different from AdaBoost and SVM classifiers at 1% level. These results mean that MGM-Boost can provide robust and stable solutions to multi-classproblems such as bond rating.

• Journal of Intelligence and Information Systems
• /
• v.16 no.4
• /
• pp.159-172
• /
• 2010

The recommender system is one of the possible solutions to assist customers in finding the items they would like to purchase. To date, a variety of recommendation techniques have been developed. One of the most successful recommendation techniques is Collaborative Filtering (CF) that has been used in a number of different applications such as recommending Web pages, movies, music, articles and products. CF identifies customers whose tastes are similar to those of a given customer, and recommends items those customers have liked in the past. Numerous CF algorithms have been developed to increase the performance of recommender systems. Broadly, there are memory-based CF algorithms, model-based CF algorithms, and hybrid CF algorithms which combine CF with content-based techniques or other recommender systems. While many researchers have focused their efforts in improving CF performance, the theoretical justification of CF algorithms is lacking. That is, we do not know many things about how CF is done. Furthermore, the relative performances of CF algorithms are known to be domain and data dependent. It is very time-consuming and expensive to implement and launce a CF recommender system, and also the system unsuited for the given domain provides customers with poor quality recommendations that make them easily annoyed. Therefore, predicting the performances of CF algorithms in advance is practically important and needed. In this study, we propose an efficient approach to predict the performance of CF. Social Network Analysis (SNA) and Artificial Neural Network (ANN) are applied to develop our prediction model. CF can be modeled as a social network in which customers are nodes and purchase relationships between customers are links. SNA facilitates an exploration of the topological properties of the network structure that are implicit in data for CF recommendations. An ANN model is developed through an analysis of network topology, such as network density, inclusiveness, clustering coefficient, network centralization, and Krackhardt's efficiency. While network density, expressed as a proportion of the maximum possible number of links, captures the density of the whole network, the clustering coefficient captures the degree to which the overall network contains localized pockets of dense connectivity. Inclusiveness refers to the number of nodes which are included within the various connected parts of the social network. Centralization reflects the extent to which connections are concentrated in a small number of nodes rather than distributed equally among all nodes. Krackhardt's efficiency characterizes how dense the social network is beyond that barely needed to keep the social group even indirectly connected to one another. We use these social network measures as input variables of the ANN model. As an output variable, we use the recommendation accuracy measured by F1-measure. In order to evaluate the effectiveness of the ANN model, sales transaction data from H department store, one of the well-known department stores in Korea, was used. Total 396 experimental samples were gathered, and we used 40%, 40%, and 20% of them, for training, test, and validation, respectively. The 5-fold cross validation was also conducted to enhance the reliability of our experiments. The input variable measuring process consists of following three steps; analysis of customer similarities, construction of a social network, and analysis of social network patterns. We used Net Miner 3 and UCINET 6.0 for SNA, and Clementine 11.1 for ANN modeling. The experiments reported that the ANN model has 92.61% estimated accuracy and 0.0049 RMSE. Thus, we can know that our prediction model helps decide whether CF is useful for a given application with certain data characteristics.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.139-156
• /
• 2021

The biggest reason for using a deep learning model in image classification is that it is possible to consider the relationship between each region by extracting each region's features from the overall information of the image. However, the CNN model may not be suitable for emotional image data without the image's regional features. To solve the difficulty of classifying emotion images, many researchers each year propose a CNN-based architecture suitable for emotion images. Studies on the relationship between color and human emotion were also conducted, and results were derived that different emotions are induced according to color. In studies using deep learning, there have been studies that apply color information to image subtraction classification. The case where the image's color information is additionally used than the case where the classification model is trained with only the image improves the accuracy of classifying image emotions. This study proposes two ways to increase the accuracy by incorporating the result value after the model classifies an image's emotion. Both methods improve accuracy by modifying the result value based on statistics using the color of the picture. When performing the test by finding the two-color combinations most distributed for all training data, the two-color combinations most distributed for each test data image were found. The result values were corrected according to the color combination distribution. This method weights the result value obtained after the model classifies an image's emotion by creating an expression based on the log function and the exponential function. Emotion6, classified into six emotions, and Artphoto classified into eight categories were used for the image data. Densenet169, Mnasnet, Resnet101, Resnet152, and Vgg19 architectures were used for the CNN model, and the performance evaluation was compared before and after applying the two-stage learning to the CNN model. Inspired by color psychology, which deals with the relationship between colors and emotions, when creating a model that classifies an image's sentiment, we studied how to improve accuracy by modifying the result values based on color. Sixteen colors were used: red, orange, yellow, green, blue, indigo, purple, turquoise, pink, magenta, brown, gray, silver, gold, white, and black. It has meaning. Using Scikit-learn's Clustering, the seven colors that are primarily distributed in the image are checked. Then, the RGB coordinate values of the colors from the image are compared with the RGB coordinate values of the 16 colors presented in the above data. That is, it was converted to the closest color. Suppose three or more color combinations are selected. In that case, too many color combinations occur, resulting in a problem in which the distribution is scattered, so a situation fewer influences the result value. Therefore, to solve this problem, two-color combinations were found and weighted to the model. Before training, the most distributed color combinations were found for all training data images. The distribution of color combinations for each class was stored in a Python dictionary format to be used during testing. During the test, the two-color combinations that are most distributed for each test data image are found. After that, we checked how the color combinations were distributed in the training data and corrected the result. We devised several equations to weight the result value from the model based on the extracted color as described above. The data set was randomly divided by 80:20, and the model was verified using 20% of the data as a test set. After splitting the remaining 80% of the data into five divisions to perform 5-fold cross-validation, the model was trained five times using different verification datasets. Finally, the performance was checked using the test dataset that was previously separated. Adam was used as the activation function, and the learning rate was set to 0.01. The training was performed as much as 20 epochs, and if the validation loss value did not decrease during five epochs of learning, the experiment was stopped. Early tapping was set to load the model with the best validation loss value. The classification accuracy was better when the extracted information using color properties was used together than the case using only the CNN architecture.

• Tuberculosis and Respiratory Diseases
• /
• v.81 no.1
• /
• pp.42-48
• /
• 2018

Background: The diffusing capacity of the lung is influenced by multiple factors such as age, sex, height, weight, ethnicity and smoking status. Although a prediction equation for the diffusing capacity of Korea was proposed in the mid-1980s, this equation is not used currently. The aim of this study was to develop a new prediction equation for the diffusing capacity for Koreans. Methods: Using the data of the Korean National Health and Nutrition Examination Survey, a total of 140 nonsmokers with normal chest X-rays were enrolled in this study. Results: Using linear regression analysis, a new predicting equation for diffusing capacity was developed. For men, the following new equations were developed: carbon monoxide diffusing capacity (DLco)=-10.4433-0.1434${\times}$age (year)+0.2482${\times}$heights (cm); DLco/alveolar volume (VA)=6.01507-0.02374${\times}$age (year)-0.00233${\times}$heights (cm). For women the prediction equations were described as followed: DLco=-12.8895-0.0532${\times}$age (year)+0.2145${\times}$heights (cm) and DLco/VA=7.69516-0.02219${\times}$age (year)-0.01377${\times}$heights (cm). All equations were internally validated by k-fold cross validation method. Conclusion: In this study, we developed new prediction equations for the diffusing capacity of the lungs of Koreans. A further study is needed to validate the new predicting equation for diffusing capacity.