Ausam/doc/man/man6/regr.6

.th REGR VI 19/10/77
.sh NAME
regr \*- multiple linear regresion program
.sh SYNOPSIS
.bd regr
input file
.sh DESCRIPTION
.it Regr
is a multiple linear regression program written in fortran for
the A.G.S.M. at U.N.S.W.. It has extremely rigid input requirements
with respect to the format of the data in 'input file'.It is necessary
to have the data set out thus:
.s3
.nf
.in 8
x1,x2,x3,x4,x5
.br
x6,x7,x8,x9,x10
.br
x11,......
.fi
.in
.s3
It is imperative that there be 
.it at
.it most
five data values per line.If the number of data values per observation
is not a multiple of five,then the residual data values should
be placed on the following line.
.lp +6 5

E.g.

If a particular regression has thirteen variables per observation
,then the data for observation n would look like this:
.s3
.nf
.in 12
x1,x2,x3,x4,x5
.br
x6,x7,x8,x9,x10
.br
x11,x12,x13
.fi
.in
.s3
.i0
.dt
(
For those interested the format is 5f14.5)
.s3
.bd Note
that there are no commas after the last data value on each
line of the input file.

.bd Note
also that each observation should start on a new line .I.e. do 
.bd not
begin observation n+1 on the last line of observation n. (This
will cause the irrecoverable loss of parts of your data and generally
kludge the results.)
.s3
The major limitations of the program are as follows:
.br
	1) Maximum no. of observations per regression :
32767
.br
	2) Maximum no. of variables (dependent included)
: 40
.br
(This limitation can be changed if it is deemed absolutely necessary.)
.br
	3) Maximum no. of regressions on any one data base
: 32767.
.s3
.sh FILES
Fort13	temporary file for residual analysis(removed at
end of run)
Residuals	
.sh "SEE ALSO"
\'\'Fortran Guide\'\' For runtime diagnostics
.sh DIAGNOSTICS
\'end of data\' self explanatory.
Any other errors will generally be unrecoverable fortran runtime
errors,caused by such things as input conversion (a character
was typed where a digit was expected).
.sh BUGS
Being written in fortran is this program's greatest failing! However
all the necessary subroutines had already been written and it
seemed pointless and effortful to reconstruct them in a decent
language.