Help System (web edition)

/LMSELECT <data>                                         / K.Vehkalahti
computes linear regression model selection statistics for all combinations
of models specified with active variables in <data>. The activations are
indicated using the MASK specification. The regressand is activated by Y
and the regressors by X's. Other variables must be set passive (by '-').

The number of models is 2^k, where k is the number of regressors.
For example, with 10 regressors there are 2^10=1024 different models.
Hence, a practical upper limit for the number of regressors is about 15.

Models include the constant term.

The following statistics of each model are saved in data file #LMSEL#:
       p = the number of parameters in the model (including constant)
     SSE = residual sum of squares
      s2 = residual variance
      R2 = coefficient of determination (R^2)
     R2a = adjusted coefficient of determination
      Cp = Mallow's Cp
     AIC = Akaike's information criterion
    SBIC = Schwarz's  bayesian information criterion
  Coeff0 = regression coefficient of the constant term
  Coeff* = regression coefficients of the regressors (*=1,2,...,9,A,B,...)
 Status* = regressor status (1=included, 0=excluded) (*=1,2,...,9,A,B,...)

The combinations of the model equations are formed with COMB operation
(see COMB?), and the parameters are estimated by ESTIMATE operation
(see ESTIMATE?).

References:

Akaike, H. (1974). A new look at the statistical model identification.
    IEEE Transactions on Automatic Control, AC-19, 716-723.
Draper, N., & Smith, H. (1998). Applied Regression Analysis, 3rd ed.,
    John Wiley & Sons.
Mallows, C. L. (1973). Some comments on Cp. Technometrics, 15, 661-676.
Ryan, T. P. (1997). Modern Regression Methods, John Wiley & Sons.

  R = More information on regression analysis