Objective The work was designed to estimate the relative importance of each variable in multiple linear model.Method The idea of the proportion of explained variance was applied and the real data were analyzed in linear model.Result The proportion of variance explained by each independent variable was different in various ordering.Conclusion The index ρ ?2 can be used to estimate the relative importance of the independent variable in a known ordering.However when an independent variable does not have a relevant ordering,it is necessary to rescue the idea by averaging relative importance over all orderings of the independent variable.
参考文献:
[1] Sander G, James J S, Michael H C. The fallacy of employing standardized regression coefficients and correlations as measures of effect, 1986(2) [2] Wen S W, Kramer M S. Etiologic fraction analysis for continuously distributed outcome variables and empirical analogy with dichotomized outcome variables. 1995(2). doi:10.1093/ije/24.2.457 [3] Kruskal W. Relative importance by averaging over ordering. 1987(1). doi:10.2307/2684310 [4] Morrison D F. Applied linear statistical methods, 1983 [5] 金丕焕. 医用统计方法, 1993 [6] 郭祖超. 医用数理统计方法, 1988