• Communications for Statistical Applications and Methods
• /
• v.11 no.3
• /
• pp.553-562
• /
• 2004

In this short communication we discuss the bias problems of CART in split variable selection and suggest a method to reduce the variable selection bias. Penalties proportional to the number of categories or distinct values are applied to the splitting criteria of CART. The results of empirical comparisons show that the proposed modification of CART reduces the bias in variable selection.

• Communications for Statistical Applications and Methods
• /
• v.10 no.3
• /
• pp.627-635
• /
• 2003

In this short communication we discuss the bias problem of C4.5 in split variable selection and suggest a method to reduce the variable selection bias among categorical predictor variables. A penalty proportional to the number of categories is applied to the splitting criterion gain of C4.5. The results of empirical comparisons show that the proposed modification of C4.5 reduces the size of classification trees.

• The Korean Journal of Applied Statistics
• /
• v.17 no.3
• /
• pp.459-473
• /
• 2004

It has well known that an exhaustive search algorithm suggested by Breiman et. a1.(1984) has a trend to select the variable having relatively many possible splits as an splitting rule. We propose an algorithm to overcome this variable selection bias problem and then construct unbiased regression trees based on the algorithm. The proposed algorithm runs two steps of selecting a split variable and determining a split rule for binary split based on the split variable. Simulation studies were performed to compare the proposed algorithm with Breiman et a1.(1984)'s CART(Classification and Regression Tree) in terms of degree of variable selection bias, variable selection power, and MSE(Mean Squared Error). Also, we illustrate the proposed algorithm with real data sets.

• The Korean Journal of Applied Statistics
• /
• v.17 no.2
• /
• pp.347-357
• /
• 2004

It is very important to select a split variable in constructing the classification tree. The efficiency of a classification tree algorithm can be evaluated by the variable selection bias and the variable selection power. The C4.5 has largely biased variable selection due to the influence of many distinct values in variable selection and the QUEST has low variable selection power when a continuous predictor variable doesn't deviate from normal distribution. In this thesis, we propose the SRT algorithm which overcomes the drawback of the C4.5 and the QUEST. Simulations were performed to compare the SRT with the C4.5 and the QUEST. As a result, the SRT is characterized with low biased variable selection and robust variable selection power.

• Journal of the Korean Data and Information Science Society
• /
• v.16 no.2
• /
• pp.195-205
• /
• 2005

In decision tree analysis, C4.5 and CART algorithm have some problems of computational complexity and bias on variable selection. But QUEST algorithm solves these problems by dividing the step of variable selection and split point selection. When input variables are continuous, QUEST algorithm uses ANOVA F-test under the assumption of normality and homogeneity of variances. In this paper, we investigate the influence of violation of normality assumption and effect of the transformation of variables in the QUEST algorithm. In the simulation study, we obtained the empirical powers of variable selection and the empirical bias of variable selection after transformation of variables having various type of underlying distributions.

• Communications for Statistical Applications and Methods
• /
• v.9 no.3
• /
• pp.809-824
• /
• 2002

we propose two methods which separate the variable selection step and the split-point selection step. We call these two algorithms as CHITES method and F&CHITES method. They adapted some of the best characteristics of CART, CHAID, and QUEST. In the first step the variable, which is most significant to predict the target class values, is selected. In the second step, the exhaustive search method is applied to find the splitting point based on the selected variable in the first step. We compared the proposed methods, CART, and QUEST in terms of variable selection bias and power, error rates, and training times. The proposed methods are not only unbiased in the null case, but also powerful for selecting correct variables in non-null cases.

• Proceedings of the Korean Statistical Society Conference
• /
• 2001.11a
• /
• pp.89-94
• /
• 2001

In this study we focus on variable selection in decision tree growing structure. Some of the splitting rules and variable selection algorithms are discussed. We propose a competitive variable selection method based on Kruskal-Wallis test, which is a nonparametric version of ANOVA F-test. Through a Monte Carlo study we note that CART has serious bias in variable selection towards categorical variables having many values, and also QUEST using F-test is not so powerful to select informative variables under heavy tailed distributions.

• The Korean Journal of Applied Statistics
• /
• v.14 no.2
• /
• pp.475-486
• /
• 2001

데이터마이닝 패키지에 구현된 분류나무 알고리즘 가운데 CART, CHAID, QUEST, C4.5에서 변수 선택법을 비교하였다. CART의 전체탐색법이 편의를 갖는다는 사실은 잘알려졌으며, 여기서는 상품화된 패키지들에서 이들 알고리즘의 편의와 선택력을 모의실험 연구를 통하여 비교하였다. 상용 패키지로는 CART, Enterprise Miner, AnswerTree, Clementine을 사용하였다. 본 논문의 제한된 모의실험 연구 결과에 의하면 C4.5와 CART는 모두 변수선택에서 심각한 편의를 갖고 있으며, CHAID와 QUEST는 비교적 안정된 결과를 보여주고 있었다.

• Korean Medical Education Review
• /
• v.12 no.1
• /
• pp.3-8
• /
• 2010

This study is designed to evaluate the effectiveness of teaching or studying programs, and thus to overcome the selectionbias in studies. Selection-bias derived from unobservable characteristics in the course of participants selection of the teaching or studying programs, in the case of cross-section data instrumental variable(IV) method and two stage least square estimation were suggested as an analysis tool. Panel data were analyzed by using both fixed effect in which individual effects are captured by intercept terms and random effect estimation where an unobserved effect can be characterized as being randomly drawn from a given distribution.

• Communications for Statistical Applications and Methods
• /
• v.22 no.1
• /
• pp.41-54
• /
• 2015

In this article, we propose nonconcave penalties on a reduced-rank regression model to select variables and estimate coefficients simultaneously. We apply HARD (hard thresholding) and SCAD (smoothly clipped absolute deviation) symmetric penalty functions with singularities at the origin, and bounded by a constant to reduce bias. In our simulation study and real data analysis, the new method is compared with an existing variable selection method using $L_1$ penalty that exhibits competitive performance in prediction and variable selection. Instead of using only one type of penalty function, we use two or three penalty functions simultaneously and take advantages of various types of penalty functions together to select relevant predictors and estimation to improve the overall performance of model fitting.