A modification of Efroymson's technique for stepwise regression analysis
暂无分享,去创建一个
The computational technique conventionally used for stepwise multiple linear regression requires the storage of an n × n matrix of data. When the number of variables, n, is large, this requirement taxes the storage capacity of presently used machinery. The near symmetry of the matrices involved permits a modification requiring only half the storage and computations of the conventional algorithm and this additional storage allows the analysis of problems containing more variables. Alternatively, it permits the analysis of problems containing the same number of variables but with all computations performed in double precision.
[1] M. J. Garside,et al. The Best Sub‐Set in Multiple Regression Analysis , 1965 .
[2] H. Breaux. ON STEPWISE MULTIPLE LINEAR REGRESSION , 1967 .
[3] M. Kendall,et al. The discarding of variables in multivariate analysis. , 1967, Biometrika.
[4] J. Oosterhoff. On the selection of independent variables in a regression equation : Preliminary report , 1963 .
[5] Gary Lotto,et al. On the Generation of All Possible Stepwise Combinations , 1962 .