FitUtil - Curve Fitting Utility APL+Win Workspace

OVERVIEW

General Utility Workspace for Curve Fitting.

DESCRIPTION

The main functions are FIT and TFIT.  Both will perform
10 different curve fittings on a set of X and Y values.

The function TFIT will removed the non-significatives
points by performing a Student test on the data.

The results are stored in a global nested variable called
∆FIT or ∆TFIT depending of the main function used.

FIT and TFIT will present the results of the curve fittings
in the APL session, the function SHOW will present them
graphically using ∆FIT or ∆TFIT. It uses QPLOT developed
by Jeff Koloseus and the APL2000 grid.

The usage of the functions is as follow:

x FIT y
SHOW ∆FIT

x TFIT y
SHOW ∆TFIT

The regressions performed are:

1. Linear:      Y' = a + b×X
2. Exponential: Y' = a × exp(b×X)
3. Power:       Y' = a × X*b
4. Logarithmic: Y' = a + b×ln(X)
5. Parabolic:   Y' = a + b×X + c×X*2
6. Cubic:       Y' = a + b×X + c×X*2 + d×X*3
7. Linear through origin:    Y' = 0 + b×X
8. Parabolic through origin: Y' = 0 + b×X + c×X*2
9. Cubic through origin:     Y' = 0 + b×X + c×X*2 + d×X*3
10.Weibull:     Y' = a × exp(b×X*c)

The first 9 curve fittings are a least square fit using the
APL's domino operator.  The Weibull regression is obtained by
iterations and may not converge to a solution all the time.

EXAMPLES

Here is an example of what you can do:

      (⍳10) FIT 2+4×⍳10

FitUtilDemo.png

VERSION

May 2008        - First Release of WS

DOWNLOAD

The workspace in APL+Win version 5 can be downloaded here: FitUtil.w3

INFORMATION

Original author:

Pierre Gilbert

Responsible:

PierreGilbert

Email:

<apgil AT SPAMFREE videotron DOT ca>


CategoryApl2000

apwFitUtil (last edited 2015-04-04 12:25:54 by PierreGilbert)