결측치가 있는 실험계획 데이터의 분석 방법에 관한 소고

Abstract

A data-oriented approach is proposed to analyze experiments designed data with missing values resulting from the invalidity of the design points. The procedure consists of representing available data in a normal probability plot, identifying better design points and the factor levels common to these points, and then deciding on optimal conditions in the experimental region or the direction of improvement. Proposed approach is illustrated with analyses of a fractional factorial design and a robust design data. The procedure is expected to be useful to the practitioners implementing design of experiments to product/process development.