GAUSS ROUTINES

WARNING: THIS PAGE IS NOT MAINTAINED ANYMORE. I STOPPED USING GAUSS IN 2010

Addel.g = deletion diagnostics for linear or quadratic discriminant analysisaddt.g = t test for additional explanatory variable

addvar.g = added variable plot using constructed variable coming

from transformation

addvarw.g = added variable plot using additional real variable

andrews.g = Andrews' curves

anova.g = to test equality of multivariate means

boxplot.g = univariate boxplot

boxplotb.g = to superimpose a bivariate boxplot on a scatter diagram

conflr.g = sign sqrt of lik.
ratio for transformations (expansion around a set of values of
\lambda)

conflrad.g = sign sqrt of
lik. ratio for transformations in discriminant analysis

elms.g = enumerate all subsamples
(n choose p)

Sampling without replacement

elmsr.g = enumerate all
subsamples

Sampling with replacement

eqvar.g = to test homogeneity of
covariances in different groups

flr.g = forward version of the likelihood ratio test for transformation

flrld.g = forward version of the likelihood ratio test for transformation in linear discriminant analysis

flrqd.g = forward version of the likelihood ratio test for transformation in quadratic discriminant analysis

fwdbsb.g = in each step of the forward search in multivariate analysis the units forming subset are stored

fldell.g = plots confidence ellipses in selected steps of the forward search

fwdglm.g = forward search for Generalized Linear Models

fwdlda.g = forward search in linear discriminant analysis

fwdmle.g = estimates of the transformation parameters in each step of the forward search

fwdmleld.g = estimates of the transformation parameters in each step of the forward search in linear discriminant analysis

fwdmleqd.g = estimates of the transformation parameters in each step of the forward search in quadratic discriminant analysis

fwdmles.g = estimates of the transformation parameter in each step of the forward search imposing a common value of lambda for all the variables

fwdols.g = forward search in regression

fwdolsmdr.g = simplified version of fwdols.g. This routine returns only the forward estimates of the minimum deletion residual, s^2 and the regression coefficients

fwdolsst.g= estimates of the forward deletion t-statistics

fwdpca.g= Forward search in principal component analysis

fwdqda.g= Forward search in quadratic discriminant analysis

glm.g = to fit a generalized linear model

glmdel = deletion diagnostics for generalized linear models

hull.g = convex hull peeling

inputbox.g= Compute necessary values to create a univariate boxplot

lda.g = linear discriminant analysis

ldasimpl.g = simplified version of lda.g

likla.g = likelihood and score test for different values of the

transformation parameter
λ
in linear regression models

liklabs.g = likelihood and score test for different values of the

transformation parameter \lambda (both sides of the equation

are transformed) in linear regression models

liklag.g = to calculate the score test for different values

of the transformation parameter \lambda in linear regression models.

Both response and explanatory can be transformed

lms.g = to compute least median of squares
(or least trimmed of aquares) estimator

lmsbs.g = to calculate the least median of squares estimator when

both sides of a model are transformed

lmsg.g = to compute the least median of squares estimator when both

response and explanatory are transformed

lmsglm.g = least median of squared in generalized linear models

lmsnls.g = least median of squares in non linear regression models

lraddel.g = deletion diagnostic based on likelihood ratio test for

transformation parameters in linear and quadratic

discriminant analysis

lrdel.g = deletion diagnostic based on likelihood ratio test for

transformation parameters in multivariate analysis

medb.g = to produce univariate or bivariate medians

multout.g = to monitor particular distances (e.g. max. distance inside subset, min. distance outside subset) in each step of the forward search

multsimp.g = simplified
version of routine multout.g

nls.g = non linear least
squares

norm.g = to test multivariate
normality

outc.g = to detect the units
which lie outside a B-spline curve

pca.g = principal component
analysis

predglm.g = to produce
reiduals in generalized linear models

given an input vector of beta coefficients

prednls.g = to produce
residuals in non linear models given

an input vector of beta coefficients

qda.g = quadratic discriminant
analysis

qdasimpl.g = simplified
version of qda.g

qqnorm.g = qqnorm plot

quelplot.g = to produce
the inputs to draw a bivariate ellipse

regressi.g = linear
regression models

rflr.g = fwd search for lik.
ratio for transformation in multivariate analysis

H0:
λ=λ_{0}
when matrix Y has a regression structure

rfwdmle.g = fwd serach for
maximum likelihood estimates of

transformation parameters of the columns of a data matrix Y, when

Y has a regression structure

rfwdmles.g = fdw search
for a common maximum likelihood estimate of a transformation

parameter of a data matrix Y, when Y has a regression structure

rob.g
= robust methods for estimating regression coefficients

using routine optmum

scatter.g = scatter plot
matrix with univariate boxplots on the main diagonal

It also enables you to specifiy different groups

scatterb.g = scatter plot
matrix with superimposed bivariate boxplots.

It also enables you to specifiy different groups

scglm.g = goodness of link
test in generalized linear models

scodel.g = deletion
diagnostic for transformations in linear regression models

scom.g = multivariate version
of the score test for linear regression models.

The additional variables are costructed automatically from
transformation

parameters vector λ

scomR.g = multivariate
version of the score test for linear regression models

The additional variables are supplied by the user

simenv.g = simulation
envelopes for qqplots

splinem.g = to superimpose
a B-spline curve on a polygon

stand.g = to standardize the data

unibiv.g = to detect
univariate and bivariate outliers from a multivariate data matrix.

It superimposes robust ellipses in each scatter diagram and counts
the number of times units fall outside the outer contuors for each
pair of variables.

vardec.g = to decompose
total deviance inside groups and between groups

vcxm.g = max lik. var-covar
matrix from data matrix

wilks.g = deletion
diagnostics in multivariate analysis using ratio

between determinants

without.g =sample without
replacement