[Analogue-commits] r287 - in pkg: . R inst man tests/Examples
noreply at r-forge.r-project.org
noreply at r-forge.r-project.org
Sun Sep 9 13:42:43 CEST 2012
Author: gsimpson
Date: 2012-09-09 13:42:42 +0200 (Sun, 09 Sep 2012)
New Revision: 287
Modified:
pkg/NAMESPACE
pkg/R/stdError.R
pkg/inst/ChangeLog
pkg/man/stdError.Rd
pkg/tests/Examples/analogue-Ex.Rout.save
Log:
properly sort out the weighted standard deviation, plus adds print method and new computation function to avoid repeating code. Allows a weighted or unweighted SD to be used.
Modified: pkg/NAMESPACE
===================================================================
--- pkg/NAMESPACE 2012-09-07 22:47:59 UTC (rev 286)
+++ pkg/NAMESPACE 2012-09-09 11:42:42 UTC (rev 287)
@@ -199,6 +199,7 @@
S3method(print, residuals.bootstrap.mat)
S3method(print, residuals.mat)
S3method(print, roc)
+S3method(print, stdError)
S3method(print, summary.analog)
S3method(print, summary.bootstrap.mat)
S3method(print, summary.cma)
Modified: pkg/R/stdError.R
===================================================================
--- pkg/R/stdError.R 2012-09-07 22:47:59 UTC (rev 286)
+++ pkg/R/stdError.R 2012-09-09 11:42:42 UTC (rev 287)
@@ -6,49 +6,61 @@
UseMethod("stdError")
}
-`stdError.mat` <- function(object, k, ...) {
- getOrd <- function(dis) {
- nas <- is.na(dis)
- order(dis[!nas])
- }
- getWts <- function(i, dis, ords, k.seq) {
- nas <- is.na(dis[,i])
- dis[!nas, i][ords[,i]][k.seq]
- }
- getEnv <- function(i, dis, ords, k.seq, y){
- nas <- is.na(dis[,i])
- y[!nas][ords[,i]][k.seq]
- }
+`stdError.mat` <- function(object, k, weighted = FALSE, ...) {
if(missing(k)) {
- k <- getK(object, ...)
- auto <- object$auto
+ k <- getK(object, weighted = weighted, ...)
} else {
- auto <- FALSE
+ attr(k, "weighted") <- weighted
+ attr(k, "auto") <- FALSE
}
- ## create k sequence
- k.seq <- seq_len(k)
- ## ordering of objects in terms of dissim
- ords <- apply(object$Dij, 2, getOrd)
- SEQ <- seq_len(ncol(ords))
- ## weights = 1/Dij
- wi <- 1 / sapply(SEQ, getWts, object$Dij, ords, k.seq, USE.NAMES = FALSE)
- ## produce matrix of Env data for each site
- env <- sapply(SEQ, getEnv, object$Dij, ords, k.seq,
- object$orig.y, USE.NAMES = FALSE)
- ## mean of env of k closest analogues
- ybar <- colMeans(env)
- ## sum weights
- sum.wi <- colSums(wi)
- sum.wi2 <- colSums(wi^2)
- sum2.wi <- sum.wi^2
- frac <- sum.wi / (sum2.wi - sum.wi2)
- wtdSD <- sqrt(frac * colSums(wi * sweep(env, 2, ybar, "-")^2))
+ wtdSD <- .stdError(object$Dij, k, object$orig.y, weighted = weighted)
names(wtdSD) <- names(object$orig.y)
class(wtdSD) <- "stdError"
+ attr(wtdSD, "k") <- k
+ attr(wtdSD, "weighted") <- attr(k, "weighted")
+ attr(wtdSD, "auto") <- attr(k, "auto")
wtdSD
}
-`stdError.predict.mat` <- function(object, k, ...) {
+`stdError.predict.mat` <- function(object, k, weighted = FALSE, ...) {
+ if(missing(k)) {
+ k <- getK(object, weighted = weighted, ...)
+ } else {
+ attr(k, "weighted") <- weighted
+ attr(k, "auto") <- FALSE
+ }
+ wtdSD <- .stdError(object$Dij, k, object$observed, weighted = weighted)
+ names(wtdSD) <- colnames(object$predictions$model$predicted)
+ class(wtdSD) <- "stdError"
+ attr(wtdSD, "k") <- k
+ attr(wtdSD, "weighted") <- attr(k, "weighted")
+ attr(wtdSD, "auto") <- attr(k, "auto")
+ wtdSD
+}
+
+`print.stdError` <- function(x, digits = min(4, getOption("digits")),
+ ...) {
+ wtd <- attr(x, "weighted")
+ cat("\n")
+ writeLines(strwrap(paste(ifelse(wtd, "Weighted standard", "Standard"),
+ "deviation of MAT predictions"),
+ prefix = "\t"))
+ cat("\n")
+ writeLines(paste(" k-analogues:", attr(x, "k")))
+ writeLines(paste(" Weighted :", wtd))
+ cat("\n")
+ nams <- attr(x, "names")
+ attributes(x) <- NULL
+ names(x) <- nams
+ print.default(zapsmall(x), digits = digits, ...)
+}
+
+##' Interal computation function for the weighted SD
+##'
+##' @param dis matrix; dissimilarity matrix in full matrix form
+##' @param k numeric; the number of analogues to use
+##' @param y numeric; vector of observed responses
+.stdError <- function(dis, k, y, weighted = FALSE) {
getOrd <- function(dis) {
nas <- is.na(dis)
order(dis[!nas])
@@ -61,33 +73,28 @@
nas <- is.na(dis[,i])
y[!nas][ords[,i]][k.seq]
}
- if(missing(k)) {
- k <- getK(object, ...)
- auto <- object$auto
- } else {
- auto <- FALSE
- }
## create k sequence
k.seq <- seq_len(k)
## ordering of objects in terms of dissim
- ords <- apply(object$Dij, 2, getOrd)
+ ords <- apply(dis, 2, getOrd)
SEQ <- seq_len(ncol(ords))
- ## weights = 1/Dij
- wi <- 1 / sapply(SEQ, getWts, object$Dij, ords, k.seq, USE.NAMES = FALSE)
## produce matrix of Env data for each site
- env <- sapply(SEQ, getEnv, object$Dij, ords, k.seq,
- object$observed, USE.NAMES = FALSE)
- ## mean of env of k closest analogues
- ybar <- colMeans(env)
- ## sum weights
- sum.wi <- colSums(wi)
- sum.wi2 <- colSums(wi^2)
- sum2.wi <- sum.wi^2
- frac <- sum.wi / (sum2.wi - sum.wi2)
- wtdSD <- sqrt(frac * colSums(wi * sweep(env, 2, ybar, "-")^2))
- names(wtdSD) <- colnames(object$predictions$model$predicted)
- class(wtdSD) <- "stdError"
- attr(wtdSD, "k") <- k
- attr(wtdSD, "auto") <- object$auto
- wtdSD
+ env <- sapply(SEQ, getEnv, dis, ords, k.seq,
+ y, USE.NAMES = FALSE)
+ res <- if(weighted) {
+ ## weights = 1/Dij
+ wi <- 1 / sapply(SEQ, getWts, dis, ords, k.seq, USE.NAMES = FALSE)
+ ## sum weights
+ sum.wi <- colSums(wi)
+ sum.wi2 <- colSums(wi^2)
+ sum2.wi <- sum.wi^2
+ frac <- sum.wi / (sum2.wi - sum.wi2)
+ ## weighted mean of env of k closest analogues
+ ybar <- colSums(env * wi) / sum.wi ## colMeans(env)
+ ## weighted standard deviation for weights not summing to 1
+ sqrt(frac * colSums(wi * sweep(env, 2, ybar, "-")^2))
+ } else {
+ apply(env, 2, sd)
+ }
+ res
}
Modified: pkg/inst/ChangeLog
===================================================================
--- pkg/inst/ChangeLog 2012-09-07 22:47:59 UTC (rev 286)
+++ pkg/inst/ChangeLog 2012-09-09 11:42:42 UTC (rev 287)
@@ -2,8 +2,19 @@
Version 0.9-10
- * stdError: calculation assumed weights summed to 1.
+ * stdError: Several changes and enhancements:
+ Calculation assumed weights summed to 1. New formula as
+ described in Simpson (2012) is now used. (Reported by Steve
+ Juggins)
+
+ Now have a choice whether to use the weighted SD or not. For
+ predictions based on the mean of the k-closest analogues it
+ would be odd to then use a weighted SD to compute the standard
+ error.
+
+ Gained a print method.
+
* caterpillarPlot: can now be called by the shorter name
caterpillar().
Modified: pkg/man/stdError.Rd
===================================================================
--- pkg/man/stdError.Rd 2012-09-07 22:47:59 UTC (rev 286)
+++ pkg/man/stdError.Rd 2012-09-09 11:42:42 UTC (rev 287)
@@ -4,16 +4,17 @@
\alias{stdError.predict.mat}
\title{Standard error of MAT fitted and predicted values}
\description{
- Computes the weighted standard deviation of the environment for the
- \emph{k}-closest analogues for each sample. This measure is proposed
- as a measure of reconstruction uncertainty for MAT models.
+ Computes the (weighted) standard deviation of the environment for the
+ \emph{k}-closest analogues for each sample. This was proposed as one
+ measure of reconstruction uncertainty for MAT models (ter Braak,
+ 1995).
}
\usage{
stdError(object, ...)
-\method{stdError}{mat}(object, k, ...)
+\method{stdError}{mat}(object, k, weighted = FALSE, ...)
-\method{stdError}{predict.mat}(object, k, ...)
+\method{stdError}{predict.mat}(object, k, weighted = FALSE, ...)
}
\arguments{
\item{object}{Object for which the uncertainty measure is to be
@@ -21,12 +22,20 @@
\code{\link{predict.mat}}.}
\item{k}{numeric; how many analogues to take? If missing, the default,
\code{k} is chosen using \code{\link{getK}}.}
+ \item{weighted}{logical; use a weighted computation?}
\item{\dots}{Additional arguments passed to other methods. Currently
not used.}
}
-%\details{
-% TODO
-%}
+\details{
+ Two types of standard error can be produced depending upon whether the
+ mean or weighted mean of \eqn{y} for the \eqn{k} closest analogues is
+ used for the MAT predictions. If \code{weighted = FALSE} then the
+ usual standard deviation of the response for the \eqn{k} closest
+ analogues is returned, whereas for \code{weighted = TRUE} a weighted
+ standard deviation is used. The weights are the inverse of the
+ dissimilarity between the target observation and each of the \eqn{k}
+ closest analogues.
+}
\value{
A named numeric vector of weighted standard deviations of the
environment for the \emph{k} closest analogues used to compute the MAT
@@ -36,11 +45,20 @@
indicating the number of analogues used and whether this was
determined from \code{object} or supplied by the user.
}
-%\references{ ~put references to the literature/web site here ~ }
+\references{
+ Simpson, G.L. (2012) Analogue methods in palaeolimnology. In Birks,
+ H.J.B, Lotter, A.F. Juggins S., and Smol, J.P. (Eds) \emph{Tracking
+ Environmental Change Using Lake Sediments, Volume 5: Data Handling and
+ Numerical Techniques}. Springer, Dordrecht.
+
+ ter Braak, C.J.F. (1995) Non-linear methods for multivariate
+ statistical calibration and their use in palaeoecology: a comparison
+ of inverse (\emph{k}-nearest neighbours, partial least squares, and
+ weighted averaging partial least squares) and classical
+ approaches. \emph{Chemometrics and Intelligent Laboratory Systems}
+ \strong{28}:165--180.
+}
\author{Gavin L. Simpson}
-%\note{ ~~further notes~~
-%
-%}
\seealso{\code{\link{minDC}}, \code{\link{mat}},
\code{\link{predict.mat}}.}
\examples{
@@ -59,9 +77,15 @@
## fit the MAT model using the squared chord distance measure
ik.mat <- mat(ImbrieKipp, SumSST, method = "SQchord")
-## standard errors
+## standard errors - unweighted
stdError(ik.mat)
+## standard errors - weighted version for above
+stdError(ik.mat, k = getK(ik.mat), weighted = TRUE)
+## standard errors - weighted; note this uses more (7) analogues
+## than the above as this model had lowest LOO error
+stdError(ik.mat, weighted = TRUE)
+
## reconstruct for the V12-122 core data
coreV12.mat <- predict(ik.mat, V12.122, k = 3)
## standard errors
Modified: pkg/tests/Examples/analogue-Ex.Rout.save
===================================================================
--- pkg/tests/Examples/analogue-Ex.Rout.save 2012-09-07 22:47:59 UTC (rev 286)
+++ pkg/tests/Examples/analogue-Ex.Rout.save 2012-09-09 11:42:42 UTC (rev 287)
@@ -1,5 +1,5 @@
-R version 2.15.1 Patched (2012-08-29 r60485) -- "Roasted Marshmallows"
+R version 2.15.1 Patched (2012-07-27 r60002) -- "Roasted Marshmallows"
Copyright (C) 2012 The R Foundation for Statistical Computing
ISBN 3-900051-07-0
Platform: x86_64-unknown-linux-gnu (64-bit)
@@ -30,7 +30,7 @@
Loading required package: MASS
Loading required package: princurve
Loading required package: mgcv
-This is mgcv 1.7-19. For overview type 'help("mgcv-package")'.
+This is mgcv 1.7-6. For overview type 'help("mgcv-package")'.
This is analogue 0.9-10
>
> assign(".oldSearch", search(), pos = 'CheckExEnv')
@@ -4950,25 +4950,25 @@
wa> ## residuals for the training set
wa> residuals(mod4)
V14.61 V17.196 V18.110 V16.227 V14.47 V23.22
--3.89845110 -0.95914236 -0.57577610 0.83646773 -1.12654865 1.85855684
+-3.89845110 -0.95914237 -0.57577610 0.83646773 -1.12654866 1.85855684
V2.12 V23.29 V12.43 R9.7 A157.3 V23.81
- 4.93907447 -1.56332718 1.02901346 -2.01091601 0.72379237 -2.30397582
+ 4.93907447 -1.56332718 1.02901346 -2.01091600 0.72379238 -2.30397581
V23.82 V12.53 V23.83 V12.56 A152.84 V16.50
--1.16893886 -1.90464621 0.99721685 -1.39764003 1.29346294 -0.29234573
+-1.16893885 -1.90464620 0.99721686 -1.39764003 1.29346295 -0.29234573
V22.122 V16.41 V4.32 V12.66 V19.245 V4.8
- 2.13000055 -4.44031821 -0.92783839 0.99172654 -1.49152688 1.88724027
+ 2.13000056 -4.44031822 -0.92783840 0.99172654 -1.49152689 1.88724026
A180.15 V18.34 V20.213 V19.222 A180.39 V16.189
- 2.78850049 -0.63727536 0.33475038 -0.01275751 -1.65993108 -2.09930701
+ 2.78850048 -0.63727536 0.33475037 -0.01275751 -1.65993109 -2.09930701
V12.18 V7.67 V17.165 V19.310 V16.190 A153.154
--0.97538682 2.31383515 1.87380536 2.64327583 0.01900072 0.13747169
+-0.97538682 2.31383514 1.87380535 2.64327583 0.01900071 0.13747169
V19.308 V22.172 V10.98 V22.219 V16.33 V22.204
--0.26612390 -2.21647192 2.40270759 0.25826513 -1.69856608 0.20738893
+-0.26612390 -2.21647192 2.40270758 0.25826513 -1.69856608 0.20738893
V20.167 V10.89 V12.79 V19.216 V14.90 A180.72
-0.92731700 -0.84954530 -0.55321548 0.80956624 0.66973330 0.70524398
V16.21 A180.76 V15.164 A180.78 V14.5 V3.128
-0.23696237 -0.14589252 -0.26826158 0.54819699 -0.32786314 1.64879299
A179.13 V9.31 V20.230 V20.7 V20.234 V18.21
- 1.53212229 0.96958483 0.44047759 -0.17038549 -0.21308146 -0.39014886
+ 1.53212230 0.96958483 0.44047759 -0.17038549 -0.21308146 -0.39014885
V12.122
0.72061292
>
@@ -6525,73 +6525,111 @@
> ## fit the MAT model using the squared chord distance measure
> ik.mat <- mat(ImbrieKipp, SumSST, method = "SQchord")
>
-> ## standard errors
+> ## standard errors - unweighted
> stdError(ik.mat)
- V14.61 V17.196 V18.110 V16.227 V14.47 V23.22 V2.12 V23.29
-3.7017647 2.7940379 2.4904961 3.5144020 2.7300757 3.0418167 2.9199864 1.0952777
- V12.43 R9.7 A157.3 V23.81 V23.82 V12.53 V23.83 V12.56
-2.4168224 3.8347560 1.6422409 1.8922602 1.9662188 2.6825814 0.5043179 2.9932954
- A152.84 V16.50 V22.122 V16.41 V4.32 V12.66 V19.245 V4.8
-3.3706748 2.2382896 2.0969686 3.0440301 0.0000000 1.0495554 2.6358967 2.1929384
- A180.15 V18.34 V20.213 V19.222 A180.39 V16.189 V12.18 V7.67
-2.9604059 1.6382546 2.5610904 2.2776091 0.9980639 0.6977540 0.1364087 1.0693862
- V17.165 V19.310 V16.190 A153.154 V19.308 V22.172 V10.98 V22.219
-1.1469297 1.0946184 2.0219976 0.6101154 0.6061351 0.5269860 1.4875516 0.5643260
- V16.33 V22.204 V20.167 V10.89 V12.79 V19.216 V14.90 A180.72
-1.5194344 1.4172128 1.3626793 1.1202298 0.5381647 1.0771118 0.8011406 0.6104648
- V16.21 A180.76 V15.164 A180.78 V14.5 V3.128 A179.13 V9.31
-0.6381386 0.3198533 0.4701962 0.7742820 0.5714590 1.4633510 0.5972338 0.3021622
- V20.230 V20.7 V20.234 V18.21 V12.122
-0.2980597 0.5809639 0.5854821 0.5723716 0.0000000
-attr(,"class")
-[1] "stdError"
+
+ Standard deviation of MAT predictions
+
+ k-analogues: 3
+ Weighted : FALSE
+
+ V14.61 V17.196 V18.110 V16.227 V14.47 V23.22 V2.12 V23.29
+ 3.3292 2.5658 2.5166 3.3292 2.5658 3.0000 2.5658 1.2583
+ V12.43 R9.7 A157.3 V23.81 V23.82 V12.53 V23.83 V12.56
+ 2.7839 3.2146 1.5000 2.0817 2.0207 2.6458 0.5000 3.0551
+ A152.84 V16.50 V22.122 V16.41 V4.32 V12.66 V19.245 V4.8
+ 3.4641 2.3629 2.1794 2.5166 0.0000 1.0408 2.2546 2.2546
+ A180.15 V18.34 V20.213 V19.222 A180.39 V16.189 V12.18 V7.67
+ 3.0551 1.5275 2.5981 2.2546 1.0000 0.6429 0.1155 1.1547
+ V17.165 V19.310 V16.190 A153.154 V19.308 V22.172 V10.98 V22.219
+ 1.1547 1.1547 2.0000 0.6429 0.6429 0.5292 1.5275 0.5774
+ V16.33 V22.204 V20.167 V10.89 V12.79 V19.216 V14.90 A180.72
+ 1.5535 1.5000 1.4434 1.1676 0.5000 1.0408 0.7638 0.5774
+ V16.21 A180.76 V15.164 A180.78 V14.5 V3.128 A179.13 V9.31
+ 0.5292 0.2887 0.4619 0.7638 0.5774 1.5044 0.5774 0.2887
+ V20.230 V20.7 V20.234 V18.21 V12.122
+ 0.2887 0.5774 0.5774 0.5774 0.0000
+> ## standard errors - weighted version for above
+> stdError(ik.mat, k = getK(ik.mat), weighted = TRUE)
+
+ Weighted standard deviation of MAT predictions
+
+ k-analogues: 3
+ Weighted : TRUE
+
+ V14.61 V17.196 V18.110 V16.227 V14.47 V23.22 V2.12 V23.29
+ 3.6236 2.6922 2.4905 3.4762 2.7156 2.9827 2.8185 1.0950
+ V12.43 R9.7 A157.3 V23.81 V23.82 V12.53 V23.83 V12.56
+ 2.3923 3.5829 1.6044 1.7100 1.7521 2.5962 0.4759 2.9924
+ A152.84 V16.50 V22.122 V16.41 V4.32 V12.66 V19.245 V4.8
+ 3.3505 2.2243 2.0883 2.7274 0.0000 1.0314 2.3222 2.1879
+ A180.15 V18.34 V20.213 V19.222 A180.39 V16.189 V12.18 V7.67
+ 2.9545 1.6056 2.5588 2.2645 0.9839 0.6838 0.1276 1.0083
+ V17.165 V19.310 V16.190 A153.154 V19.308 V22.172 V10.98 V22.219
+ 1.1467 1.0698 2.0196 0.5636 0.5709 0.5259 1.4834 0.5630
+ V16.33 V22.204 V20.167 V10.89 V12.79 V19.216 V14.90 A180.72
+ 1.5190 1.4169 1.3391 1.1180 0.5088 1.0723 0.8010 0.6063
+ V16.21 A180.76 V15.164 A180.78 V14.5 V3.128 A179.13 V9.31
+ 0.5797 0.3105 0.4611 0.7731 0.5710 1.4630 0.5954 0.3002
+ V20.230 V20.7 V20.234 V18.21 V12.122
+ 0.2970 0.5809 0.5851 0.5700 0.0000
>
+> ## standard errors - weighted; note this uses more (7) analogues
+> ## than the above as this model had lowest LOO error
+> stdError(ik.mat, weighted = TRUE)
+
+ Weighted standard deviation of MAT predictions
+
+ k-analogues: 7
+ Weighted : TRUE
+
+ V14.61 V17.196 V18.110 V16.227 V14.47 V23.22 V2.12 V23.29
+ 3.2045 3.8607 3.7093 3.8413 3.6463 3.3567 2.5745 1.8870
+ V12.43 R9.7 A157.3 V23.81 V23.82 V12.53 V23.83 V12.56
+ 2.1729 2.6576 2.6189 2.5574 2.0575 2.5081 1.9093 3.3261
+ A152.84 V16.50 V22.122 V16.41 V4.32 V12.66 V19.245 V4.8
+ 3.3846 2.1971 2.8468 2.3900 1.8478 1.9721 3.1867 2.3264
+ A180.15 V18.34 V20.213 V19.222 A180.39 V16.189 V12.18 V7.67
+ 3.0681 1.9150 1.9977 2.7098 1.9177 0.6841 0.7857 1.2002
+ V17.165 V19.310 V16.190 A153.154 V19.308 V22.172 V10.98 V22.219
+ 1.0991 1.3316 1.4314 0.5048 0.9144 0.5466 1.3659 0.6610
+ V16.33 V22.204 V20.167 V10.89 V12.79 V19.216 V14.90 A180.72
+ 1.2663 1.0450 1.2885 1.0775 1.0370 1.0816 1.0877 0.6390
+ V16.21 A180.76 V15.164 A180.78 V14.5 V3.128 A179.13 V9.31
+ 0.7313 0.4802 0.8549 0.8869 0.4255 1.0753 0.4841 0.7742
+ V20.230 V20.7 V20.234 V18.21 V12.122
+ 0.4281 0.5189 0.4316 0.5208 0.1722
+>
> ## reconstruct for the V12-122 core data
> coreV12.mat <- predict(ik.mat, V12.122, k = 3)
> ## standard errors
> stdError(coreV12.mat)
- 0 10 20 30 40 50 60
-72.3307103 0.6391695 0.6113086 1.6025626 1.3348327 1.3084087 0.8978862
- 70 80 90 100 110 120 130
- 1.2413641 1.4600559 1.3768654 0.4941674 0.2879595 1.6022391 0.2934360
- 140 150 160 170 180 190 200
- 1.7397088 0.9275737 0.7977268 0.8014461 1.1946764 0.2943270 1.7023691
- 210 220 230 240 250 260 270
- 2.2240272 0.5164777 0.6192205 1.4696838 1.3014299 1.2840133 1.5909909
- 280 290 300 310 320 330 340
- 1.8396982 0.5974174 1.2814680 1.1491139 1.5922597 1.4498985 1.2714704
- 350 360 370 380 390 400 410
- 0.7462475 1.2643008 1.2826104 0.4937233 2.2117480 0.4924661 1.4086468
- 420 430 440 450 460 470 480
- 0.2979207 0.5081370 0.3230163 0.0000000 1.4923076 1.2789171 0.2864583
- 490 500 510 520 530 540 550
- 1.1551482 0.5925977 0.5086959 0.4774772 1.2652373 0.1148866 1.0407605
- 560 570 580 590 600 610 620
- 0.2967980 0.2796920 0.4972385 1.2027792 0.2835480 0.9006515 0.4014728
- 630 640 650 660 670 680 690
- 0.2813064 0.5873118 0.7853038 0.2875564 0.4981700 2.2594850 1.2891113
- 700 710 720 730 740 750 760
- 0.9713892 1.4526717 1.5179814 1.0454000 0.4918791 0.2977649 0.2869260
- 770 780 790 800 810 820 830
- 1.4679702 1.1750746 0.5846293 0.5960886 1.7904245 0.2851860 1.4506051
- 840 850 860 870 880 890 900
- 1.0694963 0.5053518 1.3027428 1.2888246 1.5458275 2.0294194 0.7310171
- 910 920 930 940 950 960 970
- 1.0429069 1.2370137 0.2763181 0.3011330 0.4632124 0.4977645 0.2866663
- 980 990 1000 1010 1020 1030 1040
- 0.2804741 0.2868464 1.4217870 0.2919910 1.0469965 1.4342437 0.5022240
- 1050 1060 1070 1080 1090
- 1.3229458 0.2985295 1.3235215 1.7766052 1.3336282
-attr(,"class")
-[1] "stdError"
-attr(,"k")
-[1] 3
-attr(,"k")attr(,"auto")
-[1] FALSE
-attr(,"k")attr(,"weighted")
-[1] FALSE
-attr(,"auto")
-[1] FALSE
+
+ Standard deviation of MAT predictions
+
+ k-analogues: 3
+ Weighted : FALSE
+
+ 0 10 20 30 40 50 60 70 80 90 100
+0.5774 0.5774 0.5774 1.6073 1.2767 1.2767 0.8660 1.2583 1.4434 1.3229 0.5000
+ 110 120 130 140 150 160 170 180 190 200 210
+0.2887 1.6073 0.2887 1.7321 0.9292 0.8145 0.8145 1.1676 0.2887 1.7321 2.2546
+ 220 230 240 250 260 270 280 290 300 310 320
+0.5000 0.5774 1.4189 1.3229 1.2767 1.6073 1.8028 0.5774 1.2767 1.1547 1.6073
+ 330 340 350 360 370 380 390 400 410 420 430
+1.4434 1.2583 0.7506 1.2767 1.2767 0.5000 2.2546 0.5000 1.4189 0.2887 0.5000
+ 440 450 460 470 480 490 500 510 520 530 540
+0.2887 0.0000 1.4422 1.2767 0.2887 1.1547 0.5774 0.5000 0.5000 1.2767 0.1155
+ 550 560 570 580 590 600 610 620 630 640 650
+1.0408 0.2887 0.2887 0.5000 1.2767 0.2887 0.9019 0.4041 0.2887 0.5774 0.7638
+ 660 670 680 690 700 710 720 730 740 750 760
+0.2887 0.5000 2.2546 1.2583 0.9644 1.4422 1.5275 1.0408 0.5000 0.2887 0.2887
+ 770 780 790 800 810 820 830 840 850 860 870
+1.4422 1.1547 0.5774 0.5774 1.8028 0.2887 1.3229 1.0408 0.5000 1.3229 1.2767
+ 880 890 900 910 920 930 940 950 960 970 980
+1.6073 2.0000 0.7638 1.0408 1.3229 0.2887 0.2887 0.4619 0.5000 0.2887 0.2887
+ 990 1000 1010 1020 1030 1040 1050 1060 1070 1080 1090
+0.2887 1.4434 0.2887 1.0408 1.4434 0.5000 1.3229 0.2887 1.3229 1.8028 1.3229
>
>
>
@@ -7086,25 +7124,25 @@
> ## residuals for the training set
> residuals(mod4)
V14.61 V17.196 V18.110 V16.227 V14.47 V23.22
--3.89845110 -0.95914236 -0.57577610 0.83646773 -1.12654865 1.85855684
+-3.89845110 -0.95914237 -0.57577610 0.83646773 -1.12654866 1.85855684
V2.12 V23.29 V12.43 R9.7 A157.3 V23.81
- 4.93907447 -1.56332718 1.02901346 -2.01091601 0.72379237 -2.30397582
+ 4.93907447 -1.56332718 1.02901346 -2.01091600 0.72379238 -2.30397581
V23.82 V12.53 V23.83 V12.56 A152.84 V16.50
--1.16893886 -1.90464621 0.99721685 -1.39764003 1.29346294 -0.29234573
+-1.16893885 -1.90464620 0.99721686 -1.39764003 1.29346295 -0.29234573
V22.122 V16.41 V4.32 V12.66 V19.245 V4.8
- 2.13000055 -4.44031821 -0.92783839 0.99172654 -1.49152688 1.88724027
+ 2.13000056 -4.44031822 -0.92783840 0.99172654 -1.49152689 1.88724026
A180.15 V18.34 V20.213 V19.222 A180.39 V16.189
- 2.78850049 -0.63727536 0.33475038 -0.01275751 -1.65993108 -2.09930701
+ 2.78850048 -0.63727536 0.33475037 -0.01275751 -1.65993109 -2.09930701
V12.18 V7.67 V17.165 V19.310 V16.190 A153.154
--0.97538682 2.31383515 1.87380536 2.64327583 0.01900072 0.13747169
+-0.97538682 2.31383514 1.87380535 2.64327583 0.01900071 0.13747169
V19.308 V22.172 V10.98 V22.219 V16.33 V22.204
--0.26612390 -2.21647192 2.40270759 0.25826513 -1.69856608 0.20738893
+-0.26612390 -2.21647192 2.40270758 0.25826513 -1.69856608 0.20738893
V20.167 V10.89 V12.79 V19.216 V14.90 A180.72
-0.92731700 -0.84954530 -0.55321548 0.80956624 0.66973330 0.70524398
V16.21 A180.76 V15.164 A180.78 V14.5 V3.128
-0.23696237 -0.14589252 -0.26826158 0.54819699 -0.32786314 1.64879299
A179.13 V9.31 V20.230 V20.7 V20.234 V18.21
- 1.53212229 0.96958483 0.44047759 -0.17038549 -0.21308146 -0.39014886
+ 1.53212230 0.96958483 0.44047759 -0.17038549 -0.21308146 -0.39014885
V12.122
0.72061292
>
@@ -7254,7 +7292,7 @@
> ### * <FOOTER>
> ###
> cat("Time elapsed: ", proc.time() - get("ptime", pos = 'CheckExEnv'),"\n")
-Time elapsed: 18.418 0.198 19.118 0 0
+Time elapsed: 15.918 0.228 16.336 0 0
> grDevices::dev.off()
null device
1
More information about the Analogue-commits
mailing list