• Journal of the Korean Data and Information Science Society
• /
• v.24 no.3
• /
• pp.533-541
• /
• 2013

In this study, a simple principal component analysis using Lasso is proposed. This method consists of two steps. The first step is to compute principal components by the principal component analysis. The second step is to regress each principal component on the original data matrix by Lasso regression method. Each of new principal components is computed as the linear combination of original data matrix using the scaled estimated Lasso regression coefficient as the coefficients of the combination. This method leads to easily interpretable principal components with more 0 coefficients by the properties of Lasso regression models. This is because the estimator of the regression of each principal component on the original data matrix is the corresponding eigenvector. This method is applied to real and simulated data sets with the help of an R package for Lasso regression and its usefulness is demonstrated.

• The Korean Journal of Applied Statistics
• /
• v.27 no.4
• /
• pp.561-575
• /
• 2014

In this paper, we propose a penalized likelihood method to detect local spatial clusters associated with disease. The key computational algorithm is based on genlasso by Tibshirani and Taylor (2011). The proposed method has two main advantages over Kulldorff's method which is popoular to detect local spatial clusters. First, it is not needed to specify a proper cluster size a priori. Second, any type of covariate can be incorporated and, it is possible to find local spatial clusters adjusted for some demographic variables. We illustrate our proposed method using tuberculosis data from Seoul.

• Journal of the Korean Data and Information Science Society
• /
• v.24 no.6
• /
• pp.1341-1348
• /
• 2013

To solve the classification problems, various data mining techniques have been applied to database marketing, credit scoring and market forecasting. In this paper, we compare various techniques such as bagging, boosting, LASSO, random forest and support vector machine with the daily lens transaction data. The classical techniques-decision tree, logistic regression-are used too. The experiment shows that the random forest has a little smaller misclassification rate and standard error than those of other methods. The performance of the SVM is good in the sense of misclassfication rate and bad in the sense of standard error. Taking the model interpretation and computing time into consideration, we conclude that the LASSO gives the best result.

• The Korean Journal of Applied Statistics
• /
• v.28 no.1
• /
• pp.9-21
• /
• 2015

The support vector machine has been successfully applied to various classification areas due to its flexibility and a high level of classification accuracy. However, when analyzing imbalanced data with uneven class sizes, the classification accuracy of SVM may drop significantly in predicting minority class because the SVM classifiers are undesirably biased toward the majority class. The weighted $L_2$-norm SVM was developed for the analysis of imbalanced data; however, it cannot identify irrelevant input variables due to the characteristics of the ridge penalty. Therefore, we propose the weighted $L_1$-norm SVM, which uses lasso penalty to select important input variables and weights to differentiate the misclassification of data points between classes. We demonstrate the satisfactory performance of the proposed method through simulation studies and a real data analysis.

• Journal of the Korean Geographical Society
• /
• v.49 no.6
• /
• pp.917-934
• /
• 2014

Geographically weighted regression(GWR) model has been widely used to estimate spatially heterogeneous real estate prices. The GWR model, however, has some limitations of the selection of different price determinants over space and the restricted number of observations for local estimation. Alternatively, the geographically weighted LASSO(GWL) model has been recently introduced and received a growing interest. In this paper, we attempt to explore various local price determinants for the real estate by utilizing the GWL and its applicability to forecasting the real estate price. To do this, we developed the three hedonic models of OLS, GWR, and GWL focusing on the sales price of apartments in Seoul and compared those models in terms of model fit, prediction, and multicollinearity. As a result, local models appeared to be better than the global OLS on the whole, and in particular, the GWL appeared to be more explanatory and predictable than other models. Moreover, the GWL enabled to provide spatially different sets of price determinants which no multicollinearity exists. The GWL helps select the significant sets of independent variables from a high dimensional dataset, and hence will be a useful technique for large and complex spatial big data.

• Journal of the Korean Data and Information Science Society
• /
• v.24 no.1
• /
• pp.125-133
• /
• 2013

In this paper, penalized binary logistic regression models are employed as statistical models for determining the discharge of 668 patients with a chief complaint of dyspnea based on 11 blood tests results. Specifically, the ridge model based on $L^2$ penalty and the Lasso model based on $L^1$ penalty are considered in this paper. In the comparison of prediction accuracy, our models are compared with the logistic regression models with all 11 explanatory variables and the selected variables by variable selection method. The results show that the prediction accuracy of the ridge logistic regression model is the best among 4 models based on 10-fold cross-validation.

• The Korean Journal of Applied Statistics
• /
• v.34 no.6
• /
• pp.875-887
• /
• 2021

Value at Risk (VaR) is the most popular measure for market risk. In this paper, we consider the VaR estimation of portfolio consisting of a variety of assets based on multivariate copula model known as vine copula. In particular, sparse vine copula which penalizes too many parameters is considered. We show in the simulation study that sparsity indeed improves out-of-sample forecasting of VaR. Empirical analysis on 60 KOSPI stocks during the last 5 years also demonstrates that sparse vine copula outperforms regular copula model.

• Journal of Digital Contents Society
• /
• v.16 no.4
• /
• pp.605-613
• /
• 2015

This paper addresses the problem to make sparse the projection matrix in pattern recognition method. Recently, the size of computer program is often restricted in embedded systems. It is very often that developed programs include some constant data. For example, many pattern recognition programs use the projection matrix for dimension reduction. To improve the recognition performance, very high dimensional feature vectors are often extracted. In this case, the projection matrix can be very big. Recently, RSR(roated sparse regression) method[1] was proposed. This method has been proved one of the best algorithm that obtains the sparse matrix. We propose three methods to improve the RSR; outlier removal, sampling and elastic net RSR(E-RSR) in which the penalty term in RSR optimization function is replaced by that of the elastic net regression. The experimental results show that the proposed methods are very effective and improve the sparsity rate dramatically without sacrificing the recognition rate compared to the original RSR method.

• Journal of the Korean Data and Information Science Society
• /
• v.28 no.4
• /
• pp.721-731
• /
• 2017

In this study, principal component analysis methods using Lasso penalty are introduced. There are two popular methods that apply Lasso penalty to principal component analysis. The first method is to find an optimal vector of linear combination as the regression coefficient vector of regressing for each principal component on the original data matrix with Lasso penalty (elastic net penalty in general). The second method is to find an optimal vector of linear combination by minimizing the residual matrix obtained from approximating the original matrix by the singular value decomposition with Lasso penalty. In this study, we have reviewed two methods of principal components using Lasso penalty in detail, and shown that these methods have an advantage especially in applying to data sets that have more variables than cases. Also, these methods are compared in an application to a real data set using R program. More specifically, these methods are applied to the crime data in Ahamad (1967), which has more variables than cases.

• The Korean Journal of Applied Statistics
• /
• v.30 no.1
• /
• pp.41-55
• /
• 2017

This study predicts the average scores of top 150 PGA golf players on 132 PGA Tour tournaments (2013-2015) using data mining techniques and statistical analysis. This study also aims to predict the Top 10 and Top 25 best players in 4 different playoffs. Linear and nonlinear regression methods were used to predict average scores. Stepwise regression, all best subset, LASSO, ridge regression and principal component regression were used for the linear regression method. Tree, bagging, gradient boosting, neural network, random forests and KNN were used for nonlinear regression method. We found that the average score increases as fairway firmness or green height or average maximum wind speed increases. We also found that the average score decreases as the number of one-putts or scrambling variable or longest driving distance increases. All 11 different models have low prediction error when predicting the average scores of PGA Tournaments in 2015 which is not included in the training set. However, the performances of Bagging and Random Forest models are the best among all models and these two models have the highest prediction accuracy when predicting the Top 10 and Top 25 best players in 4 different playoffs.