Google
 

Trailing-Edge - PDP-10 Archives - decus_20tap2_198111 - decus/20-0026/gdata.doc
There are 2 other files named gdata.doc in the archive. Click here to see a list.
SUBROUTINE GDATA

PURPOSE
   GENERATE INDEPENDENT VARIABLES UP TO THE M-TH POWER (THE
   HIGHEST DEGREE POLYNOMIAL SPECIFIED) AND COMPUTE MEANS,
   STANDARD DEVIATIONS, AND CORRELATION COEFFICIENTS.  THIS
   SUBROUTINE IS NORMALLY CALLED BEFORE SUBROUTINES ORDER,
   MINV AND MULTR IN THE PERFORMANCE OF A POLYNOMIAL
   REGRESSION.

USAGE
   CALL GDATA (N,M,X,XBAR,STD,D,SUMSQ)

DESCRIPTION OF PARAMETERS
   N	 - NUMBER OF OBSERVATIONS.
   M	 - THE HIGHEST DEGREE POLYNOMIAL TO BE FITTED.
   X	 - INPUT MATRIX (N BY M+1) .  WHEN THE SUBROUTINE IS
	   CALLED, DATA FOR THE INDEPENDENT VARIABLE ARE
	   STORED IN THE FIRST COLUMN OF MATRIX X, AND DATA FOR
	   THE DEPENDENT VARIABLE ARE STORED IN THE LAST
	   COLUMN OF THE MATRIX.  UPON RETURNING TO THE
	   CALLING ROUTINE, GENERATED POWERS OF THE INDEPENDENT
	   VARIABLE ARE STORED IN COLUMNS 2 THROUGH M.
   XBAR  - OUTPUT VECTOR OF LENGTH M+1 CONTAINING MEANS OF
	   INDEPENDENT AND DEPENDENT VARIABLES.
   STD	 - OUTPUT VECTOR OF LENGTH M+1 CONTAINING STANDARD
	   DEVIATIONS OF INDEPENDENT AND DEPENDENT VARIABLES.
   D	 - OUTPUT MATRIX (ONLY UPPER TRIANGULAR PORTION OF THE
	   SYMMETRIC MATRIX OF M+1 BY M+1) CONTAINING CORRELA-
	   TION COEFFICIENTS.  (STORAGE MODE OF 1)
   SUMSQ - OUTPUT VECTOR OF LENGTH M+1 CONTAINING SUMS OF
	   PRODUCTS OF DEVIATIONS FROM MEANS  OF INDEPENDENT
	   AND DEPENDENT VARIABLES.

REMARKS
   N MUST BE GREATER THAN M+1.
   IF M IS EQUAL TO 5 OR GREATER, SINGLE PRECISION MAY NOT BE
   SUFFICIENT TO GIVE SATISFACTORY COMPUTATIONAL RESULTS.

SUBROUTINES AND FUNCTION SUBPROGRAMS REQUIRED
   NONE

METHOD
   REFER TO B. OSTLE, 'STATISTICS IN RESEARCH', THE IOWA STATE
   COLLEGE PRESS, 1954, CHAPTER 6.