18th edition new procedures

1. Highlights

  • 24 new procedures
  • parametric bootstrap for fixed effects in REML (VBOOTSTRAP)
  • permutation tests for random effects in REML (VRPERMTEST)
  • parallel REML analysis of several y-variates, collating the output (VAYPARALLEL)
  • space filling designs (AFDISCREPANCY, AGSPACEFILLINGDESIGN)
  • non-orthogonal multi-stratum designs (AGNONORTHOGONALDESIGN)
  • AMMI stability values
  • support vector machines (SVMFIT, SVMPREDICT)
  • VARMA models (TVARMA, TVFORECAST, TVGRAPH)
  • One-dimensional density (or violin) plots (DXDENSITY)

2. What’s new

2.1 Directives

None.

2.2 Procedures

ACONFIDENCE calculates simultaneous confidence intervals for ANOVA means.

AFDISCREPANCY calculates the discrepancy of a design.

AGNONORTHOGONALDESIGN generates non-orthogonal multi-stratum designs.

AGSPACEFILLINGDESIGN generates space filling designs.

AMTKEEP saves information from the analysis of a multitiered design by AMTIER.

AU2RDA saves results from an unbalanced analysis of variance, by AUNBALANCED, in R data frames.

DARROW adds arrows to an existing plot.

DXDENSITY produces one-dimensional density (or violin) plots.

DYPOLAR produces polar plots.

FACMERGE merges levels of factors.

FMEGAENVIRONMENTS Forms mega-environments based on winning genotypes from an AMMI-2 model.

FROWCANONICALMATRIX puts a matrix into row canonical, or reduced row echelon, form.

RRETRIEVE retrieves a regression save structure from an external file.

RSTORE stores a regression save structure in an external file.

SVMERGE merges strata prior to survey analysis.

SVMFIT fits a support vector machine.

SVMPREDICT forms the predictions using a support vector machine.

TVARMA fits a vector autoregressive moving average (VARMA) model.

TVFORECAST forecasts future values from a vector autoregressive moving average (VARMA) model.

TVGRAPH plots a vector autoregressive moving average (VARMA) model.

VAYPARALLEL does the same REML analysis for several y-variates, and collates the output.

VBOOTSTRAP performs a parametric bootstrap of the fixed effects in a REML analysis.

VRPERMTEST performs permutation tests for random terms in REML analysis.

VSPECTRALCHECK forms the spectral components from the canonical components of a multitiered design, and constrains any negative spectral components to zero.

2.2 Functions

None.

3. What’s changed

Most of the changes are compatible with Release 17, the previous release. There are a few commands, however, where new options or parameters have been inserted into the existing lists. These may cause problems in statements where option or parameter names have been omitted or abbreviated (see Section 1.7.1 of Part 1 of the Guide to the GenStat Command Language for details). To avoid any difficulty, the name of the option/parameter after the new option/parameter should be given explicitly, and not abbreviated to fewer than four characters.

Any command, where changes in Release 18 may cause incompatibilities in existing programs, is marked in Sections 3.1 and 3.2 by the symbol . The full details are given in Section 3.4.

3.1 Directives

AKEEP now adjusts degrees of freedom from repeated-measurements analyses.

CLUSTER has a new setting PRINT=random, to obtain output for the configurations obtained from every random start.

FACROTATE has two new settings of the PRINT option: loadings prints the rotated loadings (or factors), and orthogonalrotationmatrix prints the rotation matrix. The setting rotation, which was the original setting provided to print the rotated loadings, is retained as a synonym of loadings, so that existing programs will continue to run. However, in view of the confusion with the ROTATION parameter, it may be deleted in a future release.

GROUPS can now form the level or labels from the LIMITS vector, and omit the (unbounded) group that occurs below the lowest limit when BOUNDARIES=lower, or above the final limit when BOUNDARIES=upper.

REML and VKEEP now leave any restrictions on the variates supplied to save residuals or fitted values unchanged, and no longer set the values in the restricted units to missing.

3.2 Procedures

AFAUGMENTED now allows the test genotypes to have unequal replication, and the plots to contain different numbers of sub-plots.

ALLDIFFERENCES now allows you the define groupings classified by more than one factor.

AMMI now produces AMMI stability values (ASV’s), and can save the analysis-of-variance table.

AREPMEASURES now allows you to save the ANOVA save structure.

A2RDA can now save combined residuals and variates containing the treatment effect and covariate regression coefficients or the residuals (from all the strata).

DBIPLOT now allows you to control the sizes of the symbols and labels in the plot.

DIRECTORY now allows you to view subdirectories.

DFRTEXT can now add text onto rectangular plots.

DMSCATTER can now plot boxplots, density plots or dot histograms, instead of histograms, as the one-dimensional summaries.

FMFACTORS now allows you to request that the labels of multi-response factors, formed from textual codes, should remain in the order in which they are found in the data, rather than being sorted alphabetically.

FREGULAR now allows you to supply the values to insert.

GGEBIPLOT now allows you to specify the window for the key.

MACALCULATE can now save the corrected red and green values.

MAHISTOGRAM allows you to set the axis bounds.

PDESIGN can now display the factor labels in the table, instead of levels.

RWALD can now save the test probabilities and the residual degrees of freedom that are used for F statistics when the dispersion is estimated.

TABINSERT now allows you to insert values into margins of a table.

TALLY can now form tally tables for separate groupings within the data.

VAOPTIONS now allows you to request that the accumulated summary, printed from VARANDOM, should include models that fail to fit or that have bound variance parameters.

VTCOMPARISONS now allows you to specify a number of residual degrees of freedom, to use if the numbers of denominator degrees of freedom in the table of tests for fixed effects are unavailable.

3.3 Functions

None.

3.4 Incompatibilities

AFAUGMENTED procedure
option NREPTEST inserted before SEED.
DBIPLOT procedure
option SIZEMULTIPLIER inserted before SAVE.
FORMULA directive
a bug has been corrected that meant that dummies in values supplied by the VALUES parameter were not substituted, like those supplied by the VALUES option. This is reported here, although it is unlikely to affect existing programs.
SETALLOCATIONS directive
a bug meant that the default for the NREQUIRED option was saving all possible allocations, rather that just one as specified in the documentation. This has now been corrected, and so you need to set NREQUIRED to a scalar containing a missing value e.g. by NREQUIRED = !s(*) if you want to save them all.
REML and VKEEP directives
now leave any restrictions on the variates supplied to save residuals or fitted values unchanged, and no longer set the values in the restricted units to missing.
RWALD procedure
option RDF inserted before SAVE.
TALLY procedure
parameter GROUPS inserted before LIMITS.
VTCOMPARISONS procedure
options DFMETHOD, DFGIVEN and FMETHOD inserted before SAVE.