LeastSquares

1 - GENERAL FEATURES
    LeastSquares is a general purpose data analysis program
designed primarily for regression analysis linear in the
parameters. The program also has a collection of data transform,
graphics and statistics features to give the user a visual as
well as an analytical understanding of the data at hand.
    * Program written in QuickBASIC 4.5
    * Completely menu driven.
    * Graphics hardware supported: (at max resolution of each)
      > CGA, EGA, VGA, MCGA, Hercules and compatibles
      > Olivetti PC models M24,M240,M28,M280,M380,M380/C,M380/T
      > AT&T 6300 series PC's
    * Text or graphics output can be sent to IBM Graphics or
      HP LaserJet printers or disk files which can later be
      printed with the DOS print command.
    * 80x87 coprocessors are supported if present.

2 - DATA STRUCTURE
    LS can process two types of data: Columns and matrices. Data
can be read from ASCII files (with commas or blanks as
delimiters) or entered directly from the keyboard using the full
screen editor. Missing values are supported.Each column is
regarded as a variable in curve fitting. Matrices are supported
only as an auxiliary feature, mostly in relation to regression,
such as the coefficient correlation matrix etc. Since raw data
with more than 6 digit precision is extremely rare, data is
stored internally as single precision values, but all
intermediate calculations are done using double precision
arithmetic. The size of the data is limited by the memory under
640K. That means, unfortunately, that extended and/or expanded
memory is currently not supported.

3 - DATA TRANSFORMS
    * Trigonometric functions and their inverses
    * Hyperbolic functions and their inverses
    * Special functions, Legendre,Bessel etc.
    * Fourier transform and its inverse
    * Cross Correlation function (population)
    * Random # generation: Uniform, Normal and Exponential
    * Quantiles: Uniform, Normal and Exponential
    * Matrix inverse, multiply
      and more!

4 - REGRESSION
    * Simple linear and multilinear
    * Polynomial of arbitrary order
    * Weighted Fits to Orthogonal polynomials
    * Fourier series fits
    * Fits can be manually weighted
    * t-,p-values for parameters, Correlation coefficient
    * Fit plots along with data and confidence limits curve
    * Residual plots
    * Parameter spectrum
    * Save fitted values, residuals, correlation and covariance 
      matrices

5 - GRAPH
    * Scatterplot up to 5 Y-variables vs X
    * 3-D Scatterplot
    * Parametric or functional spline curves and surfaces, with
      adjustable fidelity to parent data
    * Spline and density contours
    * Histograms with density curve
    * Matrix plots
    * Completely arbitrary interactive graphical subset selection
      of data onto a clipboard.
    * User can write and draw on screen prior to printout
    * WYSIWYG screen dump of graphics to HP LaserJet or IBM     
      Graphics printers (resolution limited to that of the
screen)
    * Completely dynamic graphics: Zoom,Scroll,change ViewPt etc
      and more...

6 - UNIVARIATE STATISTICS
    * Mean, Median, StDev, Skewness, Kurtosis, Kendall's Tau,
      Percentiles etc.

7 - GENERAL USE
    If you are familiar with elementary regression, then the best
thing to do is just to go ahead and run the program with a simple
data set, such as a straight line, and try out different
options and see the results. Since it is menu driven, all
features of the program are readily accessible to the first time
user. If some of the output seems unfamiliar, such as the t- and
p-values etc., then an introductory book on statistics will be of
great help.

Since mathematical notation is essentially impossible to
implement in an ASCII file like this one, a comprehensive manual
will be available in printed form for registered users.

