[Vegan-commits] r2138 - branches/2.0/R branches/2.0/inst branches/2.0/man www
noreply at r-forge.r-project.org
noreply at r-forge.r-project.org
Sun Apr 22 08:13:39 CEST 2012
Author: jarioksa
Date: 2012-04-22 08:13:38 +0200 (Sun, 22 Apr 2012)
New Revision: 2138
Modified:
branches/2.0/R/adipart.R
branches/2.0/R/print.adipart.R
branches/2.0/inst/ChangeLog
branches/2.0/inst/NEWS.Rd
branches/2.0/man/pyrifos.Rd
www/NEWS.html
Log:
merge r2132,5 (varying gamma diversity & print in adipart) and r2137 (pyrifos.Rd edit)
Modified: branches/2.0/R/adipart.R
===================================================================
--- branches/2.0/R/adipart.R 2012-04-22 05:46:18 UTC (rev 2137)
+++ branches/2.0/R/adipart.R 2012-04-22 06:13:38 UTC (rev 2138)
@@ -63,26 +63,27 @@
index <- match.arg(index)
weights <- match.arg(weights)
switch(index,
- "richness" = {
- divfun <- function(x) apply(x > 0, 1, sum)},
- "shannon" = {
- divfun <- function(x) diversity(x, index = "shannon", MARGIN = 1, base=base)},
- "simpson" = {
- divfun <- function(x) diversity(x, index = "simpson", MARGIN = 1)})
- sumMatr <- sum(lhs)
+ "richness" = {
+ divfun <- function(x) apply(x > 0, 1, sum)},
+ "shannon" = {
+ divfun <- function(x) diversity(x, index = "shannon", MARGIN = 1, base=base)},
+ "simpson" = {
+ divfun <- function(x) diversity(x, index = "simpson", MARGIN = 1)})
## this is the function passed to oecosimu
wdivfun <- function(x) {
+ ## matrix sum *can* change in oecosimu (but default is constant sumMatr)
+ sumMatr <- sum(x)
if (fullgamma) {
tmp <- lapply(1:(nlevs-1), function(i) t(model.matrix(ftmp[[i]], rhs)) %*% x)
- tmp[[nlevs]] <- matrix(colSums(lhs), nrow = 1, ncol = ncol(lhs))
+ tmp[[nlevs]] <- matrix(colSums(x), nrow = 1, ncol = ncol(x))
} else {
tmp <- lapply(1:nlevs, function(i) t(model.matrix(ftmp[[i]], rhs)) %*% x)
}
## weights will change in oecosimu thus need to be recalculated
if (weights == "prop")
wt <- lapply(1:nlevs, function(i) apply(tmp[[i]], 1, function(z) sum(z) / sumMatr))
- else wt <- lapply(1:nlevs, function(i) rep(1 / NROW(tmp[[i]]), NROW(tmp[[i]])))
+ else wt <- lapply(1:nlevs, function(i) rep(1 / NROW(tmp[[i]]), NROW(tmp[[i]])))
a <- sapply(1:nlevs, function(i) sum(divfun(tmp[[i]]) * wt[[i]]))
if (relative)
a <- a / a[length(a)]
@@ -91,15 +92,15 @@
}
if (nsimul > 0) {
sim <- oecosimu(lhs, wdivfun, method = method, nsimul=nsimul,
- burnin=burnin, thin=thin)
- } else {
- sim <- wdivfun(lhs)
- tmp <- rep(NA, length(sim))
- sim <- list(statistic = sim,
- oecosimu = list(z = tmp, pval = tmp, method = NA, statistic = sim))
- }
+ burnin=burnin, thin=thin)
+ } else {
+ sim <- wdivfun(lhs)
+ tmp <- rep(NA, length(sim))
+ sim <- list(statistic = sim,
+ oecosimu = list(z = tmp, pval = tmp, method = NA, statistic = sim))
+ }
nam <- c(paste("alpha", 1:(nlevs-1), sep="."), "gamma",
- paste("beta", 1:(nlevs-1), sep="."))
+ paste("beta", 1:(nlevs-1), sep="."))
names(sim$statistic) <- attr(sim$oecosimu$statistic, "names") <- nam
attr(sim, "call") <- match.call()
attr(sim, "index") <- index
Modified: branches/2.0/R/print.adipart.R
===================================================================
--- branches/2.0/R/print.adipart.R 2012-04-22 05:46:18 UTC (rev 2137)
+++ branches/2.0/R/print.adipart.R 2012-04-22 06:13:38 UTC (rev 2138)
@@ -3,7 +3,11 @@
{
n <- if (is.null(x$oecosimu$simulated))
0 else ncol(x$oecosimu$simulated)
- cat("adipart with", n, "simulations\n")
+ if (n > 0)
+ cat("adipart with", n, "simulations using method",
+ dQuote(x$oecosimu$method), "\n")
+ else
+ cat("adipart ")
att <- attributes(x)
att$names <- att$call <- att$class <- att$n.levels <- att$terms <- att$model <- NULL
cat("with", paste(names(att), att, collapse=", "))
Modified: branches/2.0/inst/ChangeLog
===================================================================
--- branches/2.0/inst/ChangeLog 2012-04-22 05:46:18 UTC (rev 2137)
+++ branches/2.0/inst/ChangeLog 2012-04-22 06:13:38 UTC (rev 2138)
@@ -4,6 +4,10 @@
Version 2.0-4 (opened March 9, 2012)
+ * merge r2137: explain data transformation in pyrifos.Rd.
+ * merge r2135: print.adipart displays null model method.
+ * merge r2132: adipart bug fix: assumed constant gamma in
+ permutations.
* merge r2129: envfit failed with empty factor levels.
* merge r2128: anova(<prc-object>, by = ...) failed.
* merge r2127: more configurable msoplot.
Modified: branches/2.0/inst/NEWS.Rd
===================================================================
--- branches/2.0/inst/NEWS.Rd 2012-04-22 05:46:18 UTC (rev 2137)
+++ branches/2.0/inst/NEWS.Rd 2012-04-22 06:13:38 UTC (rev 2138)
@@ -7,6 +7,13 @@
\subsection{BUG FIXES}{
\itemize{
+ \item \code{adipart} assumed constant gamma diversity in
+ simulations when assessing the \eqn{P}-value. This could give
+ biased results if the null model produces variable gamma
+ diversities and option \code{weights = "prop"} is used. The
+ default null model (\code{"r2dtable"}) and the default option
+ (\code{weights = "unif"}) were analysed correctly.
+
\item \code{anova(<prc-object>, by = "axis")} and other
\code{by} cases failed due to \file{NAMESPACE} issues.
Modified: branches/2.0/man/pyrifos.Rd
===================================================================
--- branches/2.0/man/pyrifos.Rd 2012-04-22 05:46:18 UTC (rev 2137)
+++ branches/2.0/man/pyrifos.Rd 2012-04-22 06:13:38 UTC (rev 2138)
@@ -10,7 +10,7 @@
}
\usage{data(pyrifos)}
\format{
- A data frame with 132 observations on the log-transformed abundances
+ A data frame with 132 observations on the log-transformed (\code{log(10*x + 1)}) abundances
of 178 species. There are only twelve sites (ditches, mesocosms), but
these were studied repeatedly in eleven occasions. The treatment
levels, treatment times, or ditch ID's are not in the data frame, but
Modified: www/NEWS.html
===================================================================
--- www/NEWS.html 2012-04-22 05:46:18 UTC (rev 2137)
+++ www/NEWS.html 2012-04-22 06:13:38 UTC (rev 2138)
@@ -18,13 +18,21 @@
<ul>
+<li> <p><code>adipart</code> assumed constant gamma diversity in
+simulations when assessing the <i>P</i>-value. This could give
+biased results if the null model produces variable gamma
+diversities and option <code>weights = "prop"</code> is used. The
+default null model (<code>"r2dtable"</code>) and the default option
+(<code>weights = "unif"</code>) were analysed correctly.
+</p>
+</li>
<li> <p><code>anova(<prc-object>, by = "axis")</code> and other
<code>by</code> cases failed due to ‘<span class="file">NAMESPACE</span>’ issues.
</p>
</li>
<li> <p><code>envfit</code> failed with unused factor levels.
</p>
-</li></ul>
+</ul>
@@ -37,7 +45,7 @@
<li> <p><code>msoplot</code> is more configurable, and allows, for
instance, setting y-axis limits.
</p>
-</li></ul>
+</ul>
@@ -74,7 +82,7 @@
<a href="https://github.com/jarioksa/vegan">GitHub</a>
by Eduard Szöcs (Uni Landau, Germany).
</p>
-</li></ul>
+</ul>
@@ -106,7 +114,7 @@
was reported as an error in <code>ordiplot</code> in
<a href="https://stat.ethz.ch/pipermail/r-sig-ecology/2012-February/002768.html">R-sig-ecology</a> mailing list.
</p>
-</li></ul>
+</ul>
@@ -147,7 +155,7 @@
intensity. Thanks to consultation to Yong Cao (Univ Illinois,
USA).
</p>
-</li></ul>
+</ul>
@@ -189,7 +197,7 @@
speed-up in the internal f.test() function is fully
realised. Reported by Nicholas Lewin-Koh.
</p>
-</li></ul>
+</ul>
@@ -209,7 +217,7 @@
are class centroids of each point, but for constrained ordination
with no <code>groups</code> they are the LC scores.
</p>
-</li></ul>
+</ul>
@@ -250,7 +258,7 @@
[<a href="http://www.esajournals.org/doi/abs/10.1890/ES10-00117.1">doi:10.1890/ES10-00117.1</a>],
and was developed with the consultation of Brian Inouye.
</p>
-</li></ul>
+</ul>
@@ -289,7 +297,7 @@
results but you probably wish to upgrade <span class="pkg">vegan</span> to avoid
annoying warnings.
</p>
-</li></ul>
+</ul>
@@ -304,7 +312,7 @@
showed that in most cases an improved ordering was found rather
early in tries, and the results are equally good in most cases.
</p>
-</li></ul>
+</ul>
@@ -340,7 +348,7 @@
<span class="pkg">permute</span>, but currently only <code>betadisper</code> uses the new
feature.
</p>
-</li></ul>
+</ul>
@@ -396,7 +404,7 @@
were implemented because they were found good for species-area
models by Dengler (<EM>J. Biogeogr.</EM> 36, 728-744; 2009).
</p>
-</li></ul>
+</ul>
@@ -438,7 +446,7 @@
the length of the <code>pool</code> matches the number of
observations.
</p>
-</li></ul>
+</ul>
@@ -457,7 +465,7 @@
was not intended for normal use, but packages depending on that
code in <span class="pkg">vegan</span> should instead depend on <span class="pkg">permute</span>.
</p>
-</li></ul>
+</ul>
@@ -470,7 +478,7 @@
<li> <p><code>treeheight</code> uses much snappier code. The results
should be unchanged.
</p>
-</li></ul>
+</ul>
More information about the Vegan-commits
mailing list