MATRIX FIT

MATRIX FIT (LET)

Let Subcommand

Compute linear least squares fits for each column (or row) of a matrix with an independent variable.

This command will perform a linear least squares fit of each column (or row) of a matrix with the same independent variable. It returns the intercept and slope parameters with their corresponding standard deviations.

LET <a0> <a0sd> <a1> <a1sd> = MATRIX COLUMN FIT <mat> <x>
                        <SUBSET/EXCEPT/FOR qualification>
where <mat> is a matrix for which the fits are to be generated;
            <x> is a response variable that contains the independent variable for the fit;
            <a0> is a variable where the intercepts from the fits are saved;
            <a0sd> is a variable where the standard deviations of the intercepts from the fits are saved;
            <a1> is a variable where the slopes from the fits are saved;
            <a1sd> is a variable where the standard deviations of the slopes from the fits are saved;
and where the <SUBSET/EXCEPT/FOR qualification> is optional and rarely used in this context.

With this syntax, each column of <mat> is used as the dependent variable in a linear least squares fit with <x> as the independent variable. The number of rows in <mat> must be the same as the number of rows in <x>

LET <a0> <a0sd> <a1> <a1sd> = MATRIX ROW FIT <mat> <x>
                        <SUBSET/EXCEPT/FOR qualification>
where <mat> is a matrix for which the fits are to be generated;
            <x> is a response variable that contains the independent variable for the fit;
            <a0> is a variable where the intercepts from the fits are saved;
            <a0sd> is a variable where the standard deviations of the intercepts from the fits are saved;
            <a1> is a variable where the slopes from the fits are saved;
            <a1sd> is a variable where the standard deviations of the slopes from the fits are saved;
and where the <SUBSET/EXCEPT/FOR qualification> is optional and rarely used in this context.

With this syntax, each row of <mat> is used as the dependent variable in a linear least squares fit with <x> as the independent variable. The number of columns in <mat> must be the same as the number of rows in <x>

LET A0 A0SD A1 A1SD = MATRIX COLUMN FIT M X

None

FIT	= Perform a linear or non-linear least squares fit.
LINEAR SLOPE	= Compute the slope from a linear least squares fit.
LINEAR INTERCEPT	= Compute the intercept from a linear least squares fit.

Exploratory Data Analysis

2010/07

 
dimension 500 columns
skip 25
read gear.dat y tag
.
let tagdist = distinct tag
let tagdist = sort tagdist
let ngroup = size tagdist
.
loop for k = 1 1 ngroup
    let aval = tagdist(k)
    let y^k  = y
    retain y^k  subset tag = aval
end of loop
.
let nrow = 10
let x = sequence 1 1 nrow
let m = create matrix y1 y2 y3 y4 y5 y6 y7 y8 y9 y10
set write decimals 5
.
let b0 b1 b0sd b1sd = matrix column fit m x
.
print b0 b1 b0sd b1sd
pause
.
let mt = matrix transpose m
let c0 c1 c0sd c1sd = matrix row fit mt x
print c0 c1 c0sd c1sd