부등제한조건이 있는 정규 선형모형에서 Bayes factor를 이용한 모형선택

Abstract

부등제한조건이 있는 정규선형 회귀모형에서의 추론은 응용된 계량경제학에서 가장 흔하게 다루어지는 것 중의 하나이다. 이것은 보통, 경제학 모형은 설명변수의 부호를 제한하는 경우가 많고, 정규선형모형이매우 편리한 통계학적 도구이기 때문이다. 이러한 제한이 있는 경우의모형선택의 문제는 베이지안 해법을 통해서 쉽게 해결할 수가 있다. 가설검증 또는 모형선택에 대한 베이지안 접근방법은 Bayes factor를 이용하는 것이다. Bayes factor는 자료로부터 제공되는 모형간의 비교를위해서 사용되고 분포이론이 불필요하므로 실제 분포를 모르는 경우에도 유용하게 사용된다. 이때, Bayes factor는 복잡한 수치적 적분을 요구하는데, 이것은 컴퓨터를 이용한 몬테카를로 수치적분을 통한 시뮬레이션으로 비교적 쉽게 해결할 수가 있다. 본 논문에서는 프로그래밍 언어로 포트란 4.0을 사용하였다. ; Inference in the inequality constrained normal linear regression model is one of the most common tasks in applied economics. It is usually undertaken because the economic model restricts the signs of explanatory variables and the normal linear regression model is a very convinient statistical tool. This problem can be solved easily by bayesian solution. The bayesian approach to model selection use bayes factor. Bayes factor is used for comparison among the model and do not need density theory. Therefore it can be used easily when we do not know the true density function. The numerical integration of bayes factor is very complicate but we can do that easily by montecarlo method usig computer simulation. In this paper, faortran 4.0 is used for programing language of example. @newtempeh549044 The present study is primarily concerned with exploring the development of causative and passive lexical grammatical morphemes. One hundred and eighty normal children aged 4-9 years participated in the study. Each of the production and comprehension of causative and passive sentence was tested with forty and thirty-two sentences, respectively. In the comprehension test, the subject was required to select one of four pictures, and in the production test, the subject was required to orally describe the picture. The results were analyzed with ANOVA, frequencies analysis. The following conclusions were obtained (1) significant differences were found between ages, (2) development patterns of causative and passive were found to be in the order of causative production, passive comprehension, passive production, and causative comprehension, (3) the most difficult grammatical morpheme in production of causative and passive was gi , (4) error types in the production of causative and passive were developed progressively in accordance with the increase of ages.