Importance parameters (\psi) are no longer calculated when
non-parametric response functions are used.
0.3.0
Importance parameters (\psi) are enabled again when non-parametric
response functions are used, but not used for plotting.
2D sample plots for constrained ordination with non-parametric
response functions have been disabled, as they are not
interpretable. Variable plots are the only 2D plots still allowed
Explained deviance and inertia can be plotted on the axes rahter
than the (\psi)’s using the “plotPsi” argument to the plot.RCM()
function.
Possibility to provide lower dimensional fits has been disabled.
RCM is fast enough to fit the whole model.
1.0.0
Release on BioConductor
1.0.1
Bug fix in buildCovMat() to avoid false warning
Check for alias structure in confounder and covariate matrices
1.2.0
Missing values in count matrix are now allowed. They simply do not
contribute to the parameter estimation, but the rest of the row (or
column) is still used.
1.2.1
Vertical reference line in residual plot
Bug fix for problematic variable names
1.2.2
Moving the online manual information to the vignette
1.2.3
Rename a and b to rowExp and colExp to avoid partial
matching
Allow rowExp and colExp to be adapted for constrained
correspondence analysis starting values as well
1.2.4
Adding a new inflVar variable to disambiguate in the influence
plotting
More argument checking + tests for the plot.RCM function
1.5.2
Avoid returning nulls for residualPlot
1.5.4
A note in the vignette and in the help file of plot.RCM regarding
limited number of combinations of constraining variables. Also a
warning is now thrown
1.5.5
Bug fix for higher dimension residualPlot function, and tests for
this function
1.5.6
Replace deprecated guides( =FALSE) by guides(=“none”)
1.5.7
Update vignette to number table of contents
1.11.0
Added FAQ section in vignette with first frequent question on number
of samples not shown.
Fixed bugs for plots of data with missing values, and added tests.
1.11.2
Explicitly import stats::model.matrix, and only load necessary VGAM
functions
1.11.3
For the unconstrained models: fit feature models one by one and
Gram-Schmidt orthogonalize and center afterwards, rather than using
Lagrange multipliers and huge Jacobian matrices. This will use less
memory and speed up computations, but may yield slightly different
solutions. Nothing changes for the constrained models.
1.11.4
Introduction of permanova testing for user-supplied groups using the permanova function.