SPM's
The primary SPM (statistical parameter map) generated by the NPAIRS system is the mean Z-score volume, aka mean rSPM(z), created by averaging the individual Z-score volumes across the NPAIRS splits. A mean Z-score volume is computed for each dimension within each pattern type.
Continuing our example started in Running NPAIRS, we can display the mean
Z-score volume associated with the CVA canonical eigenimage by issuing
the following IDL command:
IDL> npairs_disp_vols, 'ex1_8FO', DIR='example', SELPATT='nwcgis', REFVOL='data/refVol/ref_vol_T1_128x128x48.mri'
The displayed images are the slices from the mean Z-score volume associated with the 1st CVA canonical eigenimage. The 1st parameter is the NPAIRS ID, the 2nd parameter (keyword DIR) is the NPAIRS directory, the 3rd parameter (keyword SELPATT) tells the program to look at the canonical eigenimage, and the 4th paramater (keyword REFVOL) is the name of the reference MRI volume. Note, the image displayed above has been modified by altering parameters within the survey_vols_w GUI, and thus will look different from the initial display created by invoking the above IDL command.
To display the mean Z-score volume associated with the 2nd contrast vector
from the GLM c*beta image, execute the following IDL command:
IDL> npairs_disp_vols, 'ex1_8FO', DIR='example', SELPATT='cbeta', SELDIM=2, REFVOL='data/refVol/ref_vol_T1_128x128x48.mri'
The SelPatt='cbeta' selects c*beta as the pattern type to be displayed. The SelDim=2 refers to the second contrast vector in the GLM, which was defined as a linear ramp on the indicator columns of the design matrix, and tested for any linear time effects in the data. Thus, what is displayed is the mean Z-score volume associated with the 2nd contrast vector for the c*beta pattern type. To view the mean Z-score volume associated with the 1st contrast vector (baseline/activation) and the T-statistic pattern type, we would use SelDim=1 and SelPatt='tstat'.
The following command performs random field theory (RFT) on the mean SPM{Z} pattern. In this example RFT is applied to the mean SPM{Z} created from the c*beta patterns of a GLM NPAIRS.
IDL> npairs_spm_rf, 'ex1_8FO', [1,1,1], DIR='example', SELPATT='cbeta', DISPVOL=1, CLUSTERSIZE=5, REFVOL='data/refVol/ref_vol_T1_128x128x48.mri' Processing: ID=ex1_8FO_DEMO MODEL=GLM PATTERN=cbeta DIM=1 ....done **************************************************************************************************************************** ID: ex1_8FO_DEMO DIR: example/ MODEL: GLM PATTERN: cbeta DIM: 1 **************************************************************************************************************************** Date = Fri Jun 15 09:24:52 2001 Operator = jra Current directory = /home/stokes/jra/distrib/NPAIRS-1.0 Volume directory = example/ex1_8FO_DEMO Volume name = ex1_8FO_DEMO.GLM.SUMM.cbeta.ZS-AVG 1 Volume size = 128, 128, 48 Search Volume 1 ============================================================================================================================ set-level cluster-level voxel-level volume location ----------- ---------------------- ----------------------------- ----------------------------- C P(C) K P(K) (uncorr) Z (Z) P(Z) (uncorr) i,j,k {vox} x,y,z {mm} ============================================================================================================================ 14 0.000 631 0.000 (0.000) 9.49 ( 9.49) 0.000 (0.000) 75 58 38 35 -21 47 5.75 ( 5.75) 0.000 (0.000) 61 54 37 -7 -9 43 5.74 ( 5.74) 0.000 (0.000) 66 53 38 7 -6 47 5.52 ( 5.52) 0.001 (0.000) 71 66 37 23 -46 43 4.65 ( 4.65) 0.047 (0.000) 74 68 36 32 -53 40 4.48 ( 4.48) 0.095 (0.000) 63 55 42 -1 -12 60 3.60 ( 3.60) 0.891 (0.000) 66 59 40 7 -25 53 3.36 ( 3.36) 0.990 (0.000) 61 57 41 -7 -18 57 334 0.000 (0.000) 7.43 ( 7.43) 0.000 (0.000) 57 69 18 -20 -56 -20 6.19 ( 6.19) 0.000 (0.000) 61 69 20 -7 -56 -13 3.65 ( 3.65) 0.859 (0.000) 51 66 14 -39 -46 -33 130 0.000 (0.000) 5.96 ( 5.96) 0.000 (0.000) 58 70 11 -17 -59 -43 4.65 ( 4.65) 0.047 (0.000) 53 71 11 -32 -62 -43 3.62 ( 3.62) 0.880 (0.000) 56 64 11 -23 -40 -43 3.31 ( 3.31) 0.994 (0.000) 57 65 13 -20 -43 -37 80 0.000 (0.000) 5.56 ( 5.56) 0.001 (0.000) 80 62 28 51 -34 13 92 0.000 (0.000) 5.05 ( 5.05) 0.008 (0.000) 66 56 26 7 -15 6 3.92 ( 3.92) 0.550 (0.000) 71 54 27 23 -9 10 3.79 ( 3.79) 0.705 (0.000) 61 56 25 -7 -15 3 3.26 ( 3.26) 0.997 (0.001) 60 54 28 -10 -9 13 60 0.001 (0.000) 4.41 ( 4.41) 0.124 (0.000) 72 70 18 26 -59 -20 3.89 ( 3.89) 0.589 (0.000) 74 68 15 32 -53 -30 3.70 ( 3.70) 0.811 (0.000) 76 70 17 39 -59 -23 16 0.187 (0.023) 4.30 ( 4.30) 0.185 (0.000) 48 55 36 -48 -12 40 15 0.216 (0.027) 4.23 ( 4.23) 0.229 (0.000) 73 67 11 29 -50 -43 9 0.500 (0.076) 4.10 ( 4.10) 0.345 (0.000) 65 61 15 4 -31 -30 8 0.569 (0.092) 3.96 ( 3.96) 0.502 (0.000) 64 71 14 1 -62 -33 11 0.381 (0.052) 3.81 ( 3.81) 0.683 (0.000) 47 61 28 -51 -31 13 11 0.381 (0.052) 3.64 ( 3.64) 0.865 (0.000) 79 55 35 48 -12 37 3.44 ( 3.44) 0.974 (0.000) 79 52 35 48 -3 37 10 0.437 (0.063) 3.53 ( 3.53) 0.937 (0.000) 56 52 24 -23 -3 0 8 0.569 (0.092) 3.50 ( 3.50) 0.951 (0.000) 56 49 27 -23 6 10 ============================================================================================================================ Height threshold: Z = 3.09, p = 0.001 (1.000 corrected) Degrees of freedom = -1 Extent threshold: k = 5 voxels, p = 0.175 (0.798 corrected) Smoothness FWHM = 10.0 10.0 10.0 {mm} = 3.2 3.2 3.0 {voxels} Expected voxels per cluster,= 2.885 Search volume: S = 996780 mm^3 = 30160 voxels = 873.1 resels Expected number of clusters, = 1.600 Voxel size: 3.1, 3.1, 3.4 mm (1 resel = 30.26 voxels) ------------------- P(C) = probability of getting C or more clusters of size 5, or more, above the threshold 3.09. P(K) = probability of getting a cluster of size K, or more, above the threshold 3.09. P(Z) = probability of getting a z above Z. *********************************************************************************************** * DISPLAYING VOLUME(S) * *********************************************************************************************** -------------------------------------------------------------------- Display ID Directory Model Pattern Dim SPM -------------------------------------------------------------------- YES ex1_8FO example/ GLM cbeta 1 .ZS-AVG -------------------------------------------------------------------- NOTE: Volumes having no "significant" clusters are NOT displayed.
The second argument ([1,1,1]) is the resolution of the data in the x, y, z dimensions measured as FWHM in CM's. It is assumed FWHM is known. The SELPATT=cbeta selects the c*beta pattern type, and the DISPVOL=1 flag informs the program to display the "significant" clusters of the SPM{Z}. If this flag is not set then only the RFT table is created. By default, a significant cluster is one that has either a signficant peak value or a significant cluster size. The CLUSTERSIZE=5 indicates that clusters of size less than 5 are to be discarded.
[Top]
[SPM's]
[Reproducibility]
[Generalization]
[Subject Influence]
Reproducibility
The reproducibility of the spatial patterns generated by NPAIRS is measured by the correlation coefficient.
The first half of a split will produce a set of spatial patterns, where the number of spatial patterns
depends on the number of dimensions and the
number of pattern types. The second half
of a split will produce a similar set of spatial patterns, only generated from an independent data set. The
corresponding spatial patterns across the two splits are compared by computing their correlation coeffients.
If there are N splits of the data, then N correlation coeffients are computed for each kind of spatial pattern.
For example, in an NPAIRS running CVA, there will be N correlation coefficients for the canonical eigenimages
associated with the first dimension, N correlation coefficients for the 2nd dimension, etc.
Continuing with our example, to plot the correlation coefficients for the 8 class CVA we would enter:
The first parameter to npairs_plot_reprod is the ID of our NPAIRS analysis, which is defined by
the Id parameter in the
parameter file. The DIR keyword specifies the name of the directory where the output of the NPAIRS
analysis lives. This directory was defined by the Dir parameter in
the parameter file. The SELPATT keyword specifies the pattern type to look at, which in this case is the
CVA canonical eigenimages. The npairs_plot_reprod program first generates a table listing information about
each distribution plotted. In this example, there are 7 correlation coefficient distributions, one for each CVA
dimension. The NPAIRS Directory column holds the name of the output directory for each distribution, the
ID column holds the NPAIRS ID for each distribution, the Suffix column holds the filename suffix for the
pattern type associated with each distribution, the Dim
column holds the dimension for each distribution, the Plot
column specifies the plot number in a multi-plot for each distribution, and the X Value column specifies
the x axis coordinate for each distribution. This table becomes more useful when multiple plots are generated
from multiple NPAIRS analyses.
In the actual plot, the x axis is distribution number, which in this example runs across CVA dimensions,
and the y-axis is correlation coeffient.
To plot the same data using "box plots" use /BOXPLOT:
For the reproducibility of the patterns generated by GLM we have:
The SELMOD='GLM' selects all pattern types generated by GLM. There are two pattern types
(T-statistic (tstat) and
c*beta (cbeta), and two
"dimensions" (baseline/activation contrast vector and linear
ramp contrast vector, labeled dimensions 1 and 2, respectively).
To plot just the results for the T-statistic volumes you would use the SELPATT keyword:
The SELPATT='tstat' refers to the T-statistic volume. The two distributions plotted belong to the two contrast
vectors tested. To look at the c*beta distributions use SELPATT='cbeta'.
To plot the results for the contrast vector that test for baseline/activation effects you would use the
SELDIM keyword:
In this example, the two distributions are for the T-statistic volumes and the c*beta volumes,
respectively. The YRANGE keyword is used to set the y-axis plot range to be from .3 to .7.
To plot the distribution histograms for the c*beta volumes you would enter:
The histograms correspond to the correlation coeffients for the c*beta volumes generated for
each of the two contrast vectors in the GLM. The solid line refers to the baseline/activation contrast
vector, and the dotted refers to the linear ramp contrast vector. The vertial bars at the base of each
histogram mark the median value of the distributions (the bar for the dotted histogram is hard to see in such
a small plot window).
To plot the reproducibility of the first 3 CVA dimensions using colored histograms:
The SELDIM=[1,2,3] parameter informs the program to plot the distributions assocatiated with CVA
dimensions 1,2,3, and /COLORPLOT forces each histogram to have its own color.
See the documentation for npairs_plot_reprod for a full
listing of all the available plotting options.
[Top]
[SPM's]
[Reproducibility]
[Generalization]
[Subject Influence]
Generalization
Generalization refers to the ability of a model to predict the labels of volumes from a data set that is
independent of the data set used to estimate the model parameters. The data used to estimate the model
parameters is called the training set, and the data whose labels are to be predicted is called the test set.
How well (or poorly) a model performs on the test set is measured by the Generalization Error, which can be
computed using a number of different metrics (e.g, misclassification, posterior probability).
In NPAIRS each split of the data creates 2 independent data sets. The first half of the split is used as a training
set, and the second half as a test set. Model parmaters are estimated from the training set and "applied" to the
test set, resulting in a generalization error. Then the roles are reversed, the first half of the split becomes the
test set, and the second half becomes the training set. This produces a second generalization error. This process
is repeated for each of the splits, resulting in 2*N generalization error measurements, where N is the number of
NPAIRS splits.
Currently, only CVA NPAIRS produces generalization errors. Briefly, a CVA is run on the training set with canonical
variates and canonical eigenimages generated. Each volume in the test set, which is really a row in the test data
matrix, is projected onto the training canonical eigenimages (one for each dimension) to produce test canonical
variates. The mean canonical variates, one per class, for the training set are computed. If there are C classes
then there are C mean canonical variate vectors, with each vector having C-1 elements (the number of CVA dimensions
is one minus the number classes). Next, for each test volume, the test canonical variates (one for each CVA dimension)
are "compared" to the training mean canonical variates for each class. This results in C probabilities. The first
probability is the probability that the test volume belongs to the first class (label), the second probability is
the probability the test volume belongs to the second class, etc. Thus, C probabilities are computed for each test
volume. [These probabilities are really posterior probabilities since prior probabilities are used to compute them.]
From these C probabilities, a number of different generalization errors can be computed. The true class membership
for a test volume is known, and the posterior probability corresponding to this class is what we simply call - don't
be too confused - the posterior probability. That is, the posterior probability is the probability
that a test volume belongs to its true class, based on the model computed from the training set. The posterior
probability is computed for each volume in the test set, and the average is computed. Each split of the data will
then produce 2 mean posterior probabilities, and 2*N total for the entire NPAIRS analysis, where N is the number
of splits. These 2*N posterior probabilities are what is plotted by the NPAIRS plotting programs.
Another generalization metric is the misclassification error. A test volume is classified into one of the C classes
based on the class with the highest posterior probability. The classification error is 0 if the volume is correctly
classified, and 1, otherwise. This is done for each test volume, and the results are averaged to produce a
misclassification error for the test set. This is done twice for each split, resulting in 2*N misclassification errors.
Back to our example. To plot the posterior probabilities for our 8 class CVA we would enter:
As with npairs_plot_reprod, the first paremater is the ID for this analysis, and the keyword DIR
specifies the directory where the NPAIRS output files are located. Note, there is only one distribution plotted
even though the CVA has 7 dimensions (from 8 classes). This is because all the dimensions in a CVA are used to
compute the generalization error. This is unlike reproducibility where 7 correlation
coefficient distributions were plotted, one for each dimension. The distribution in this example consists of 70
posterior probabilities, with 2 values coming from each of the 35 splits of the data.
To plot the same data in histogram mode you would enter:
To plot misclassification errors, enter:
The STAT=2 parameter informs the progam to plot misclassification errors. The default is 0, which
is posterior probabilities. The other choices are: squared prediction error (STAT=1), log error
(STAT=3), and bit rate (STAT=4). See references
for a description of these generalization error metrics.
[Top]
[SPM's]
[Reproducibility]
[Generalization]
[Subject Influence]
Subject Influence
Subject influence measures how much each subject adds (or removes) from the
reproducibility of a data set. The first step in computing this metric is to generate a reference
spatial pattern. The default reference pattern is created by averaging the 2 spatial patterns with the
highest correlation coefficient. For N splits of the data there are N correlation coeffients which measure
the similarity between each of the spatial pattern pairs. The pair with the highest correlation coefficient
are averaged together to form a single reference pattern.
Next, for each split, the correlation coefficient
(r) is computed between the reference pattern and the 2 patterns generated by each split-half. The two r
values are compared. The subjects belonging to the split-half with the higher r value are identified, and
counters (1 per subject) for these subjects are incremented by 1. For an NPAIRS with N splits, the highest
value of a counter, for any one subject, is N. The lowest is 0. Subjects with higher counts tend to add to
the reproducibility of the data set, and subjects with lower counts tend to lessen the reproducibility.
To plot subject influence for our CVA NPAIRS example we would enter:
As with npairs_plot_reprod and npairs_plot_predict, the first paremater is the ID for this analysis,
the keyword DIR specifies the directory where the NPAIRS output files are located, the keyword SELMOD='CVA'
selects the pattern types from the CVA output, which in this case is only the nwcgis pattern, and the keyword
PMULTI=[0,3,3] is a multi-plot parameter indicating that there be 3 columns and 3 rows of plots per
plot window (the 0 means to start the plotting in the first column, first row). There is one plot
for each CVA dimension. Remember, each CVA dimension has its own spatial pattern, and therefore its own set of
reproducibility correlation coeffients. The y-axis is the counts for each subject (see desription above). The x-axis
are the subjects (or more specifically, scan sessions), ordered according to counts, with each subject identified
by its Scan Session
number. The 3 horizontal lines represent the mean (middle) and +/- 2 standard deviations (upper, lower) from a binomial
distribution with N trials and success probability of .5, where N is the number of splits. Note, we do not claim that
the influence counts have a binomial distribution. The lines are provided only as a reference.
To plot subject influence for the GLM results, enter:
In this example we have 4 plots. The first two plots are associated with
T-statistic volumes created using two different contrast vectors:
activation/baseline (left plot), and time effect (right plot). The last two plots are from the
c*beta volumes created using the activation/baseline contrast vector
(left plot), and the time effect contrast vector (right plot).
[Top]
[SPM's]
[Reproducibility]
[Generalization]
[Subject Influence]
IDL> npairs_plot_reprod, 'ex1_8FO', DIR='example', SELPATT='nwcgis'
Program = npairs_plot_reprod
Date = Wed May 30 14:06:22 2001
Current directory = /home/stokes/jra/work/pro
------------------------------------------------------------------------------------
N Plot X ID Model Pattern Dim NPAIRS Directory
------------------------------------------------------------------------------------
1 1 1 ex1_8FO CVA nwcgis 1 example/
2 1 2 ex1_8FO CVA nwcgis 2 example/
3 1 3 ex1_8FO CVA nwcgis 3 example/
4 1 4 ex1_8FO CVA nwcgis 4 example/
5 1 5 ex1_8FO CVA nwcgis 5 example/
6 1 6 ex1_8FO CVA nwcgis 6 example/
7 1 7 ex1_8FO CVA nwcgis 7 example/
------------------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 1
2: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 2
3: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 3
4: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 4
5: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 5
6: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 6
7: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 7
IDL> npairs_plot_reprod, 'ex1_8FO', DIR='example', SELPATT='nwcgis', /BOXPLOT
Program = npairs_plot_reprod
Date = Thu May 31 11:57:18 2001
Current directory = /home/stokes/jra/work/pro
------------------------------------------------------------------------------------
N Plot X ID Model Pattern Dim NPAIRS Directory
------------------------------------------------------------------------------------
1 1 1 ex1_8FO CVA nwcgis 1 example/
2 1 2 ex1_8FO CVA nwcgis 2 example/
3 1 3 ex1_8FO CVA nwcgis 3 example/
4 1 4 ex1_8FO CVA nwcgis 4 example/
5 1 5 ex1_8FO CVA nwcgis 5 example/
6 1 6 ex1_8FO CVA nwcgis 6 example/
7 1 7 ex1_8FO CVA nwcgis 7 example/
------------------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 1
2: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 2
3: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 3
4: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 4
5: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 5
6: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 6
7: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 7
IDL> npairs_plot_reprod, 'ex1_8FO', DIR='example', SELMOD='GLM'
Program = npairs_plot_reprod
Date = Thu May 31 12:01:47 2001
Current directory = /home/stokes/jra/work/pro
------------------------------------------------------------------------------------
N Plot X ID Model Pattern Dim NPAIRS Directory
------------------------------------------------------------------------------------
1 1 1 ex1_8FO GLM tstat 1 example/
2 1 2 ex1_8FO GLM tstat 2 example/
3 1 3 ex1_8FO GLM cbeta 1 example/
4 1 4 ex1_8FO GLM cbeta 2 example/
------------------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.GLM.SUMM.tstat.CC 1
2: example/ex1_8FO/ex1_8FO.GLM.SUMM.tstat.CC 2
3: example/ex1_8FO/ex1_8FO.GLM.SUMM.cbeta.CC 1
4: example/ex1_8FO/ex1_8FO.GLM.SUMM.cbeta.CC 2
IDL> npairs_plot_reprod, 'ex1_8FO', DIR='example', SELPATT='tstat'
Program = npairs_plot_reprod
Date = Thu May 31 12:25:32 2001
Current directory = /home/stokes/jra/work/pro
------------------------------------------------------------------------------------
N Plot X ID Model Pattern Dim NPAIRS Directory
------------------------------------------------------------------------------------
1 1 1 ex1_8FO GLM tstat 1 example/
2 1 2 ex1_8FO GLM tstat 2 example/
------------------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.GLM.SUMM.tstat.CC 1
2: example/ex1_8FO/ex1_8FO.GLM.SUMM.tstat.CC 2
IDL> npairs_plot_reprod, 'ex1_8FO', DIR='example', SELMOD='GLM', SELDIM=1, YRANGE=[.3,.7]
Program = npairs_plot_reprod
Date = Thu May 31 12:30:35 2001
Current directory = /home/stokes/jra/work/pro
------------------------------------------------------------------------------------
N Plot X ID Model Pattern Dim NPAIRS Directory
------------------------------------------------------------------------------------
1 1 1 ex1_8FO GLM tstat 1 example/
2 1 2 ex1_8FO GLM cbeta 1 example/
------------------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.GLM.SUMM.tstat.CC 1
2: example/ex1_8FO/ex1_8FO.GLM.SUMM.cbeta.CC 1
IDL> npairs_plot_reprod, 'ex1_8FO', DIR='example', SELPATT='cbeta', /HISTPLOT, XRANGE=[-.5,1]
Program = npairs_plot_reprod
Date = Thu May 31 12:36:26 2001
Current directory = /home/stokes/jra/work/pro
------------------------------------------------------------------------------------
N Plot X ID Model Pattern Dim NPAIRS Directory
------------------------------------------------------------------------------------
1 1 1 ex1_8FO GLM cbeta 1 example/
2 1 2 ex1_8FO GLM cbeta 2 example/
------------------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.GLM.SUMM.cbeta.CC 1
2: example/ex1_8FO/ex1_8FO.GLM.SUMM.cbeta.CC 2
IDL> npairs_plot_reprod, 'ex1_8FO', DIR='example', SELPATT='nwcgis', SELDIM=[1,2,3], /HISTPLOT, /COLORPLOT, XRANGE=[-.4,.8]
Program = npairs_plot_reprod
Date = Thu May 31 12:41:43 2001
Current directory = /home/stokes/jra/work/pro
------------------------------------------------------------------------------------
N Plot X ID Model Pattern Dim NPAIRS Directory
------------------------------------------------------------------------------------
1 1 1 ex1_8FO CVA nwcgis 1 example/
2 1 2 ex1_8FO CVA nwcgis 2 example/
3 1 3 ex1_8FO CVA nwcgis 3 example/
------------------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 1
2: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 2
3: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.CC 3
IDL> npairs_plot_predict, 'ex1_8FO', DIR='example'
Program = npairs_plot_predict
Date = Thu May 31 16:23:31 2001
Current directory = /home/stokes/jra/work/pro
----------------------------------------------------------------------------
N Plot X ID Model Pattern NPAIRS Directory
----------------------------------------------------------------------------
1 1 1 ex1_8FO CVA nwcgis example/
----------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.POST
IDL> npairs_plot_predict, 'ex1_8FO', DIR='example', /HISTPLOT, XRANGE=[0,.6]
Program = npairs_plot_predict
Date = Thu May 31 16:28:16 2001
Current directory = /home/stokes/jra/work/pro
----------------------------------------------------------------------------
N Plot X ID Model Pattern NPAIRS Directory
----------------------------------------------------------------------------
1 1 1 ex1_8FO CVA nwcgis example/
----------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.POST
IDL> npairs_plot_predict, 'ex1_8FO', DIR='example', STAT=2
Program = npairs_plot_predict
Date = Thu May 31 16:31:11 2001
Current directory = /home/stokes/jra/work/pro
----------------------------------------------------------------------------
N Plot X ID Model Pattern NPAIRS Directory
----------------------------------------------------------------------------
1 1 1 ex1_8FO CVA nwcgis example/
----------------------------------------------------------------------------
Name(s) of data files:
1: example/ex1_8FO/ex1_8FO.CVA.SUMM.nwcgis.POST
IDL> npairs_plot_influ, 'ex1_8FO', DIR='example', SELMOD='CVA', PMULTI=[0,3,3]
IDL> npairs_plot_influ, 'ex1_8FO', DIR='example', SELMOD='GLM', PMULTI=[0,4,1]
