help screeplot dialog: screeplot
-------------------------------------------------------------------------------
Title
[MV] screeplot -- Scree plot
Syntax
screeplot [eigvals] [, options ]
scree is a synonym for screeplot.
options description
-------------------------------------------------------------------------
Main
neigen(#) graph only largest # eigenvalues; default is
to plot all eigenvalues
Mean
mean graph horizontal line at the mean of the
eigenvalues
meanlopts(cline_options) affect rendition of the mean line
CI
ci[(ci_options)] graph confidence intervals (after pca only);
ci is a synonym for ci(asymptotic)
Plot
cline_options affect rendition of the lines connecting
points
marker_options change look of markers (color, size, etc.)
Add plots
addplot(plot) add other plots to the generated graph
Y axis, X axis, Titles, Legend, Overall
twoway_options any options other than by() documented in
[G] twoway_options
-------------------------------------------------------------------------
ci_options description
-------------------------------------------------------------------------
asymptotic compute asymptotic confidence intervals; the
default
heteroskedastic compute heteroskedastic bootstrap confidence
intervalss
homoskedastic compute homoskedastic bootstrap confidence
intervals
area_options affect the rendition of the confidence bands
table produce a table of confidence intervals
level(#) set confidence level; default is level(95)
reps(#) number of bootstrap simulations; default is
reps(200)
seed(#) random-number seed used for the bootstrap
simulations
-------------------------------------------------------------------------
Menu
Statistics > Multivariate analysis > Factor and principal component
analysis > Postestimation > Scree plot of eigenvalues
Description
screeplot produces a scree plot of the eigenvalues of a covariance or
correlation matrix.
screeplot automatically obtains the eigenvalues after estimation commands
that have eigen as one of their e(properties) and that store the
eigenvalues in the matrix e(Ev). These commands include candisc, discrim
lda, factor, factormat, pca, and pcamat. screeplot also works
automatically to plot singular values after ca and camat, canonical
correlations after canon, and eigenvalues after manova, mca, mds, mdsmat,
and mdslong.
screeplot lets you obtain a scree plot in other cases by directly
specifying eigvals, a vector containing the eigenvalues.
Options
+------+
----+ Main +-------------------------------------------------------------
neigen(#) specifies the number of eigenvalues to plot. The default is to
plot all eigenvalues.
+------+
----+ Mean +-------------------------------------------------------------
mean displays a horizontal line at the mean of the eigenvalues.
meanlopts(cline_options) affects the rendition of the mean reference line
added using the mean option; see [G] cline_options.
marker_options affect the rendition of markers drawn at the plotted
points, including their shape, size, color, and outline; see [G]
marker_options.
+----+
----+ CI +---------------------------------------------------------------
ci[(ci_options)] displays confidence intervals for the eigenvalues. The
option ci is a synonym for ci(asymptotic). The following methods for
estimating confidence intervals are available:
ci(asymptotic) specifies the asymptotic distribution of the
eigenvalues of a central Wishert distribution, the distribution
of the covariance matrix of a sample from a multivariate normal
distribution. The asymptotic theory applied to correlation
matrices is not fully correct, probably giving confidence
intervals that are somewhat too narrow.
ci(heteroskedastic) specifies a parametric bootstrap by using the
percentile method and assuming that the eigenvalues are from a
matrix that is multivariate normal with the same eigenvalues as
observed.
ci(homoskedastic) specifies a parametric bootstrap by using the
percentile method and assuming that the eigenvalues are from a
matrix that is multivariate normal with all eigenvalues equal to
the mean of the observed eigenvalues. For a PCA of a correlation
matrix, this mean is 1.
ci(area_options) affects the rendition of the confidence bands; see
[G] area_options.
ci(table) produces a table with the confidence intervals.
ci(level(#)) specifies the confidence level, as a percentage, for
confidence intervals. The default is level(95) or as set by set
level.
ci(reps(#)) specifies the number of simulations to be performed for
estimating the confidence intervals. This option is valid only
when heteroskedastic or homoskedastic is specified. The default
is reps(200).
ci(seed(str)) sets the random-number seed used for the parametric
bootstrap. Setting the seed makes sure that results are
reproducible. See set seed in [R] set seed. This option is
valid only when heteroskedastic or homoskedastic is specified.
The confidence intervals are not adjusted for "simultaneous
inference" (see [P] _mtest).
+------+
----+ Plot +-------------------------------------------------------------
cline_options affect the rendition of the lines connecting the plotted
points; see [G] cline_options.
+-----------+
----+ Add plots +--------------------------------------------------------
addplot(plot) provides a way to add other plots to the generated graph;
see [G] addplot_option.
+-----------------------------------------+
----+ Y axis, X axis, Titles, Legend, Overall +--------------------------
twoway_options are any of the options documented in [G] twoway_options,
excluding by(). These include options for titling the graph (see [G]
title_options) and for saving the graph to disk (see [G]
saving_option).
Examples
Setup
. webuse bg2
. factor bg2cost1-bg2cost6
Draw scree plot
. screeplot
Setup
. pca bg2cost1-bg2cost6
Draw scree plot with asymptotic confidence intervals
. screeplot, ci(asymptotic)
Also see
Manual: [MV] screeplot
Help: [MV] factor, [MV] pca, [MV] mds
