Accession Number : AD0646511

Title :   LEAST SQUARES CURVE FITTING BY COMPUTER: A LEAST SQUARES ITERATIVE TECHNIQUE IS DEVELOPED TO SOLVE MULTIPLE LINEAR REGRESSION PROBLEMS FOR N-DIMENSIONAL CURVE FITTING BY USE OF FORTRAN IV SUBROUTINES.

Descriptive Note : Technical memo.,

Corporate Author : NAVAL MISSILE CENTER POINT MUGU CALIF

Personal Author(s) : Archer,C. O.

Report Date : 20 JAN 1967

Pagination or Media Count : 67

Abstract : A Fortran IV subroutine (CURFIT) for curve fitting by multiple linear regression has been designed for the IBM 7094 digital computer. This subroutine may be used in the body of other programs without adding sigificantly to the execution time. The report presents the CURFIT subroutine and supplemental subroutines that prepare data for curve fitting in polynomial form, where the polynomials may be functions of one, two, or three independent variables. The mathematical theory for solving multiple linear regression problems in the least squares sense is presented. Subroutines are described for: (1) Fitting a polynomial of one, two, or three degrees of arbitrary order to data which come from some line, surface, or hypersurface. (2) Plotting functions of one or two independent variables. Examples are presented for one-, two-, and three-degree polynomial fits. Also, the report shows how the subroutines may be adapted for solution of special problems and situations. Complete flow charts and Fortran IV listings of all of the described subroutines are presented. (Author)

Descriptors :   (*CURVE FITTING, *DIGITAL COMPUTERS), (*LEAST SQUARES METHOD, CURVE FITTING), ITERATIONS, REGRESSION ANALYSIS, COMPUTER PROGRAMS, PROGRAMMING LANGUAGES, COMPILERS, FLOW CHARTING, POLYNOMIALS

Subject Categories : Statistics and Probability
      Computer Programming and Software
      Computer Hardware

Distribution Statement : APPROVED FOR PUBLIC RELEASE