hyperspec Plotting functions

Transcription

1 hperspec Plotting functions Claudia Beleites DIA Raman Spectroscop Group, Universit of Trieste/Ital (2 28) Spectroscop Imaging, IPHT, Jena/German (28 216) ÖPV, JKI, Berlin/German Chemometric Consulting and Chemometri GmbH, Wölfersheim/German June 27, 218 Reproducing the Eamples in this Vignette All spectra used in this manual are installed automaticall with hperspec. Note that some definitions are eecuted in vignette.defs, and others invisibl at the beginning of the file in order to have the code as similar as possible to interactive sessions. Contents 1 Predefined functions 2 2 Arguments for plot 3 Spectra Stacked spectra Calibration Plots, (Depth) Profiles, and Time Series Plots Calibration plots Time series and other Plots of the Tpe Intensit-over-Something Levelplot 14 6 Spectra Matri 14 7 False-Colour Maps: plotmap D plots (with rgl) 2 9 Using ggplot2 with hperspec objects 21 1 Troubleshooting No output is produced

2 11 Interactive Graphics spc.identif: finding out wavelength, intensit and spectrum map.identif: finding a spectrum in a map plot map.sel.pol: selecting spectra inside a polgon in a map plot Related functions provided b base graphics and lattice Suggested Packages latticeetra: available deldir: rgl: ggplot2 : available available available In addition tripack, and latticist are mentioned, but not used in this vignette. Preliminar Calculations For some plots of the chondro dataset, the pre-processed spectra and their cluster averages ± one standard deviation are more suitable: > chondro.preproc <- chondro - spc.fit.pol.below (chondro) > chondro.preproc <- chondro.preproc / rowmeans (chondro) > chondro.preproc <- chondro.preproc - quantile (chondro.preproc,.) > cluster.cols <- c ("dark blue", "orange", "#C22") > cluster.meansd <- aggregate (chondro.preproc, chondro$clusters, mean_pm_sd) > cluster.means <- aggregate (chondro.preproc, chondro$clusters, mean) For details about the pre-processing, please refer to the eample work flow in vignette ("chondro"), or the help? chondro. 1 Predefined functions hperspec comes with 6 major predefined plotting functions. plot main switchard for most plotting tasks levelplot hperspec has a method for lattice[? ] function levelplot plotspc plotmat plotc plotmap plots spectra plots the spectra matri calibration plot, time series, depth profile plotc is a lattice function more specialized version of levelplot for map or image plots. plotmap is a lattice function plotvoronoi more specialized version of plotmap that produces Voronoi tesselations. plotvoronoi is a lattice function 2

3 plotmap, plotvoronoi, and levelplot are lattice functions. Therefore, in loops, functions, Sweave chunks, etc. the lattice object needs to be printed eplicitl b e. g. print (plotmap (object)) (R FAQ: Wh do lattice/trellis graphics not work?). 3

4 I fl a.u. plotspc plots the spectra, i. e. the intensities $spc over the > plotspc (flu) λ nm row plotmat plots the spectra, i. e. the colour coded intensities $spc over 6 the and the row number. > plotmat (flu) λ nm I fl a.u. plotc c / (mg / l) plots an intensit over a single other data column, e. g. calibration time series depth profile > plotc (flu) levelplot plots a false colour map, defined b a formula. > levelplot (spc ~ *, chondro, aspect = "iso") Warning: Onl first wavelength is used for plotting 4

5 plotmap plotmap is a specialized version of levelplot. It uses a single value (e. g. average intensit or cluster membership) over two data columns (default $ and $) > plotmap (chondro) plotvoronoi 1 cell lacuna matri Function plotvoronoi is a special version of plotmap that produces Voronoi diagram of the hperspec object. > plotvoronoi (sample (chondro, 3), clusters ~ * ) Arguments for plot hperspec s plot method uses its second argument to determine which of the specialized plots to produce. This allows some hand abbreviations. All further arguments are handed over to the function actuall producing the plot. I fl a.u. plot (, spc ) is equivalent to plotspc (flu) > plot (flu, "spc") λ nm

6 plot (, spcmeansd )..1.2 plots mean spectrum ± 1 standard deviation > plot (chondro.preproc, "spcmeansd") plot (, spcprctile )..1.2 plots median, 16 th and 84 th percentile for each wavelength. For Gaussian distributed data, 16 th, th and 84 th percentile are equal to mean ± standard deviation. Spectroscopic data frequentl are not Gaussian distributed. The percentiles give a better idea of the true distribution. The are also less sensitive to outliers. > plot (chondro.preproc, "spcprctile") plot (, spcprctl )...3 like "spcprctl" plus th and 9 th percentile. > plot (chondro.preproc, "spcprctl") plot (, c ) 2 > plot (flu, "c") is equivalent to plotc (flu) I fl a.u c / (mg / l) 6

7 plot (, ts ) plots a time series plot > plot (laser [,, 4], "ts") equivalent to plotc (laser, spc ~ t) t / s plot (, depth ) z µm plots a depth profile plot > depth.profile <- new ("hperspec", + spc = as.matri (rnorm (2) + 1:2), + data = data.frame (z = 1 : 2), + labels = list (spc = "", + z = epression (`/` (z, mu*m)), +.wavelength = epression (lambda))) > plot (depth.profile, "depth") the same as plotc (laser, spc ~ z) row plot (, mat ) plots the spectra matri. > plot (laser, "mat") 6 Equivalent to > plotmat (laser) A lattice alternative is: 2 > levelplot (spc ~.wavelength *.row, laser) λ nm plot (, map ) is equivalent to plotmap (chondro) > plot (chondro, "map") 7

8 plot (, voronoi ) > plot (sample (chondro, 3), "voronoi") See plotvoronoi 3 Spectra plotspc offers a variet of parameters for customized plots. To plot... plotspc particular wavelength range 1 3 if onl one wavelength range is needed, the etract command (see vignette ("introduction")) is handiest: > plotspc (paracetamol [,, 7 ~ 12]) wavelengths. If wl.range alread contains indices use wl.inde = TRUE more wavelength ranges use wl.range = list (6 ~ 18, 28 ~ 31). Cut the wavelength ais appropriatel with offset = 7 > plotspc (paracetamol, + wl.range = c (3 ~ 18, 28 ~ ma), + offset = 7) If available, the package plotri[1] is used to produce the cut mark. with reversed abscissa 2 use wl.reverse = TRUE > plotspc (paracetamol, wl.reverse = TRUE )

9 I fl a.u. in different colours use col = vector.of.colours > plotspc (flu, col = matlab.dark.palette (6)) λ nm dots instead of lines 2 use lines.args = list (pch = 2, tpe = "p") > plotspc (paracetamol [,, 28 ~ 32], + lines.args = list (pch = 2, tpe = "p")) I a. u. mass spectra use lines.args = list (tpe = "h") > plot (barbiturates [[1]], lines.args = list (tpe = "h")) m u z e more spectra into an eisting plot use add = TRUE > plotspc (chondro [ 3,,]) > plotspc (chondro [3,,], add = TRUE, col = "blue")

10 Summar characteristics func ma be used to calculate summar characteristics prior to plotting. To plot e. g. the standard deviation of the spectra, use: > plotspc (chondro.preproc, func = sd) with different line at I = 2 zeroline takes a list with parameters to abline, NA suppresses the line. > plotspc (paracetamol, + zeroline = list (col = "red")) 2 adding to a spectra plot plotspc uses base graphics. After plotting the spectra, more content ma be added to the graphic b abline, lines, points, etc. > plot (laser, "spcmeansd") > abline (v = c(4.63, , 4.288, 4.391), + col = c("black", "blue", "red", "darkgreen")) λ nm 3.1 Stacked spectra stacked use stacked = TRUE > plotspc (cluster.means, + col = cluster.cols, + stacked = TRUE)

11 Stacking groups of spectra matri lacuna The spectra to be stacked can be grouped: stacked = factor. Alternativel, the name of the grouping etra data column can be used: > plot (cluster.meansd, + stacked = ".aggregate", + fill = ".aggregate", + col = cluster.cols) Manuall giving offset Stacking values can also be given manuall as numeric values in offset: > plotspc (cluster.meansd, + offset = rep (:2, each = 3), + col = rep (cluster.cols, each = 3)) Dense stacking To obtain a denser stacking: > offsets <- appl (cluster.means [[]], 2, diff) > offsets <- - appl (offsets, 1, min) > plot (cluster.means, offset = c (, cumsum (offsets)), + col = cluster.cols) Elaborate eample > offset <- appl (chondro.preproc, 2, quantile, c(.,.9)) > offset <- range (offset) > plot(chondro.preproc[1], + plot.args = list (lim = c (, 2) * offset), + lines.args = list( tpe = "n")) > offset <- (:1) * diff (offset) > for (i in 1 : 3){ + plot(chondro.preproc, "spcprctl", offset = offset [i], + col = "gra", add = TRUE) + plot (chondro.preproc [i], offset = offset [i], + col = matlab.dark.palette (3) [i], add = TRUE, + lines.args = list (lwd = 2)) + } plotspc allows fine grained customization of almost all aspects of the plot. This is possible b 11

12 giving arguments to the functions that actuall perform the plotting plot for setting up the plot area, lines for the plotting of the lines, ais for the aes, etc. The arguments for these functions should be given in lists as plot.args, lines.args, ais.args, etc. 4 Calibration Plots, (Depth) Profiles, and Time Series Plots 4.1 Calibration plots plotc I fl a.u. Intensities over concentration 6 3 Plotting the Intensities of one wavelength over the concentration for univariate calibration: > plotc (flu [,, 4]) The default is to use the first intensit onl c / (mg / l) range(i fl a.u.) Summar Intensities over concentration A function to compute a summar of the intensities before drawing can be used: 6 > plotc (flu, func = range, groups =.wavelength) If func returns more than one value, the different results are accessible b.wavelength c / (mg / l) Conditioning: plotting more traces separatel > plotc (flu [,, c (4, 44)], spc ~ c.wavelength, + ce =.3, scales = list (alternating = c(1, 1))) I fl a.u c / (mg / l) 12

13 Grouping: plot more traces in one panel 6 > plotc (flu [,, c (4, 44)], groups =.wavelength) I fl a.u c / (mg / l) I 4 nm a.u. Changing Ais Labels (and other parameters) Arguments for plot can be given to plotc: 6 > plotc (flu [,, 4], + lab = epression (I ["4 nm"] / a.u.), + lim = range (, flu$c +.1), + lim = range (, flu$spc + 1), 3 + pch = 4) c / (mg / l) I fl a.u. Adding things to the plot: customized panel function As plotc uses the lattice function plot, additions to the 2 plot must be made via the panel function: > panelcalibration <- function (,,..., clim = range (), level =.9 + panel.plot (,,...) + lm <- lm ( ~ ) 1 + panel.abline (coef (lm),...) + c <- seq (clim [1], clim [2], length.out = ) + c <- predict (lm, data.frame ( = c), + interval = "confidence", c / (mg / l) + level = level) + panel.lines (c, c [,2], col = "gra") + panel.lines (c, c [,3], col = "gra") + } > plotc (flu [,,4], panel = panelcalibration, + pch = 4, clim = c (,.3), level =.99) 4.2 Time series and other Plots of the Tpe Intensit-over-Something 13

14 Abscissae other than c 6 2 Other abscissae ma be specified b eplicitl giving the model formula: > plotc (laser [,, c(4.63, , 4.288, 4.391)], + spc ~ t, + groups =.wavelength, + tpe = "b", + col = c ("black", "blue", "red", "darkgreen")) t / s Levelplot hperspec s levelplot can use two special column names:.wavelength for the wavelengths.row for the row inde (i. e. spectrum number) in the data Besides that, it behaves eactl like levelplot. Particularl, the data is given as the second argument: levelplot > levelplot (spc ~ *, chondro) factors as z 1 cell lacuna If the colour-coded value is a factor, the displa is adjusted to this fact: > levelplot (clusters ~ *, chondro) matri Spectra Matri It is often useful to plot the spectra against an additional coordinate, e. g. the time for time series, the depth for depth profiles, etc. This can be done b plot (object, "mat"). The actual plotting is done b image, but levelplot 14

15 can produce spectra matri plots as well and these plots can be grouped or conditioned. different palette λ nm row > plot (laser, "mat", col = heat.colors (2)) is the same as > plotmat (laser, col = heat.colors (2)) different ais λ nm t / s Using a different etra data column for the ais: > plotmat (laser, = "t") alternativel, values and ais label can be given separatel. > plotmat (laser, = laser$t, lab = labels (laser, "t")) contour lines λ nm row Contour lines ma be added: > plotmat (flu, col = matlab.dark.palette (2)) > plotmat (flu, col = "white", + contour = TRUE, add = TRUE) colour-coded points: levelplot with special panel function m z u e t min > require ("latticeetra") > barb <- do.call (collapse, barbiturates[1:]) > barb <- orderwl (barb) > levelplot (spc ~.wavelength * z, barb, + panel = panel.levelplot.points, + ce =.33, col.smbol = NA, + col.regions = matlab.palette)

16 7 False-Colour Maps: plotmap plotmap is a specialized version of levelplot. The spectral intensities ma be summarized b a function before plotting (default: mean). The same scale is used for and aes (aspect = iso ). plotting map cell > plotmap (chondro) 1 lacuna matri plotting maps with other than and colour-coded factors specif the colour-coded variable, abscissa and ordinate as formula: colour.coded ~ abscissa * ordinate > plotmap (chondro, spc ~ * ) 1 cell lacuna > plotmap (chondro, clusters ~ * ) If the colour-coded variable is a factor, each level gets its own colour, and the legend is labeled accordingl. matri different palette To plot with a different palette, use argument col.regions. 1 cell lacuna > plotmap (chondro, clusters ~ *, + col.regions = cluster.cols) matri

17 Fine tuning lattice parameters The plotting of color maps is done via R package lattice (aka Trellis graphic approach), which is highl customizable. Use trellis.par.get and trellis.par.set to get/set the settings for the current graphics device. > m.theme = trellis.par.get() > names(m.theme) # note how man parameters are tunable [1] "grid.pars" "fontsize" "background" "panel.background" [] "clip" "add.line" "add.tet" "plot.polgon" [9] "bo.dot" "bo.rectangle" "bo.umbrella" "dot.line" [13] "dot.smbol" "plot.line" "plot.smbol" "reference.line" [17] "strip.background" "strip.shingle" "strip.border" "superpose.line" [21] "superpose.smbol" "superpose.polgon" "regions" "shade.colors" [2] "ais.line" "ais.tet" "ais.components" "laout.heights" [29] "laout.widths" "bo.3d" "par.lab.tet" "par.lab.tet" [33] "par.zlab.tet" "par.main.tet" "par.sub.tet" An of these parameters can be fine-tuned to produce the desired output. For eample, parameter m.theme$region is responsible for the appearance of color maps, and it contains elements $alpha and $col. B changing this parameters ou can create our own theme for plotting and pass it to the plotting function via par.settings. changed palette This plot uses a customized lattice theme. > m.theme$regions$col = grdevices::terrain.colors > plotmap (chondro, par.settings = m.theme) It is possible to persistentl (i.e. inside of the current R session) set lattice parameters, so the would appl to all further plots. This is done via a call to trellis.par.set, for eample trellis.par.set(m.theme). The current settings can be visualized via a call to show.settings() show current lattice settings superpose.smbol superpose.line strip.background strip.shingle dot.[smbol, line] > # Displa current trellis parameters > show.settings() World Hello bo.[dot, rectangle, umbrella] add.[line, tet] reference.line plot.[smbol, line] plot.shingle[plot.polgon] histogram[plot.polgon] barchart[plot.polgon] superpose.polgon regions 17

18 An overview of different color palettes, and was to create our own, can be found in the R color cheatsheet. defined wavelengths To plot a map of the average intensit at particular wavelengths use etraction: > plotmap (chondro.preproc [,, c(728, 782, 198, + 124, 1482, 77)], + col.regions = matlab.palette) Conditioning FALSE TRUE 8 > plotmap (chondro, 8 + spc ~ * >, col.regions = matlab.palette(2)) Conditioning on.wavelength plotmap automaticall applies the function in func before plotting. This defaults to the mean. In order to suppress this, use func = NULL. This allows conditioning on the wavelengths. To plot e. g. the first two score maps of a principal component analsis: > pca <- prcomp (~ spc, data = chondro.preproc$.) > scores <- decomposition (chondro, pca$, + label.wavelength = "PC", + label.spc = "score / a.u.") > plotmap (scores [,,1:2], + spc ~ * as.factor(.wavelength), + func = NULL, + col.regions = matlab.palette(2)) 18

19 Conditioning on.wavelength II Alternativel, use levelplot directl: > levelplot (spc ~ * as.factor(.wavelength), + scores [,,1:2], + aspect = "iso", + col.regions = matlab.palette(2)) Voronoi plot cell lacuna matri > plotvoronoi (sample (chondro, 3), clusters ~ *, + col.regions = matlab.palette(2)) Voronoi uses panel.voronoi from latticeetra[2]. The tesselation is calculated b default using deldir[3], but tripack[4] can also be used. tripack seems to faster in general, but ma hang with certain data sets (particularl regular grids with missing spectra as in this eample). Furthermore, it is not FOSS (free and open source software), so users are kindl asked to review tripack s license before using it. Mark missing spectra If the spectra come from a rectangular grid, missing positions can be marked with this panel function: > mark.missing <- function (,, z,...){ + panel.levelplot (,, z,...) + + miss <- epand.grid ( = unique (), = unique ()) + miss <- merge (miss, data.frame (,, TRUE), + all. = TRUE) + miss <- miss [is.na (miss[, 3]),] + panel.plot (miss [, 1], miss [, 2], pch = 4,...) + } > plotmap (sample (chondro, length(chondro) - 2), + col.regions = matlab.palette(2), + col = "black", + panel = mark.missing) Unevenl spaced measurement grid The panel function used b plotmap defaults to panel.levelplot.raster which assumes an evenl spaced measurement grid. Even if the spectra are measured on a nominall evenl spaced grid, the actual stage position ma be slightl varing due to positioning inaccurac and some manufacturers (e. g. Kaiser) record the position reported b the stage rather than the position requested b the stage control. 19

20 This leads to weird looking output with holes, and possibl wrong columns: > plotmap (uneven) The smptom of this situation are warnings about values in and/or not being equispaced; and that the output therefore ma be wrong One possibilit to obtain a correct map is using plotvoronoi instead which will construct a mosaic-like image with the respective piel areas being centered around the actuall recorded $ and $ position: > plotvoronoi (uneven) Another possibilit that underlines a point shape of the measurements is switching to latticeetra::panel.levelplot.points: > plotmap (uneven, panel = panel.levelplot.points, + ce =.7, col.smbol=na) Alternativel, the measurement raster positions can be rounded to their nominal raster, e. g.: > r <- makeraster (uneven$, start = -11., d = 1, tol =.3) > uneven$ <- r$ > r <- makeraster (uneven$, start = -4.77, d = 1, tol =.3) > uneven$ <- r$ > plotmap (uneven) D plots (with rgl) 2

21 rgl[] offers fast 3d plotting in R. As rgl s ais annotations are sometimes awkward, the ma better be set manuall: > laser <- laser [,,44.8 ~ 4.6] / 1 > laser$t <- laser$t / 36 > cols <- rep (matlab.palette (nrow (laser)), nwl (laser)) > surface3d ( = wl (laser), = laser$t, + z = laser$spc, col = cols) > aspect3d (c (1, 1,.2)) > aes3d (c ('+-', '--', 'z--')) > aes3d ('--', nticks = 2, labels= FALSE) > mtet3d ("t / h", '+-', line = 2.) > mtet3d ("lambda / nm", '--', line = 2.) > mtet3d ("", edge = 'z--', line = 2.) 9 Using ggplot2 with hperspec objects hperspec objects do not et directl support plotting with ggplot2 [6]. Nevertheless, ggplot2 graphics can easil be obtained, and qplot* equivalents to plotspc and plotmap are defined: plot spectra with as.long.df I fl a.u. 6 2 c > qplotspc (flu) + aes (colour = c) λ nm Map with ggplot2 1 spc > qplotmap (chondro) + + scale_fill_gradientn ("spc", colours = matlab.palette ()) The two special columns.wavelength and.rownames contain the wavelength ais and allow to distinguish the spectra. For more general plotting, as.long.df transforms a hperspec object into a long-form data.frame that is suitable for qplot, while as.t.df produces a data.frame where each spectrum is one column, and an additional first column gives the wavelength (see plotting mean ± sd below for an eample). Long data.frames can be ver memor consuming as the are of size nr ow nwl (ncol + 2) with 21

22 respect to the dimensions of the hperspec object. Thus, e. g. the chondro data set (2 MB) as hperspec object) needs 26 MB as long-format data.frame. It is therefore highl recommended to calculate the particular data to be plotted beforehand. Mean ± standard deviation with ggplot > qplotspc (mean (chondro)) + + geom_ribbon (aes (min = mean + sd, + ma = mean - sd, + =, group = NA), + alpha =.2, + data = as.t.df (mean_sd (chondro))) Note that qpotspc specifies aesthetics = spc and groups =.rownames, which do not have corresponding columns in the data.frame returned b as.t.df. These aesthetics must therefore be set manuall in the aesthetics definition in geom_ribbon (or an other geom_ that uses as.t.df). Otherwise, errors occur that object spc (and/or.rownames) cannot be found. Cut spectra with ggplot Cut aes can be implemented b faceting: > qplotspc (paracetamol / 1e4, + wl.range = c( min ~ 18, 28 ~ ma)) + + scale continuous (breaks = seq (, 32, )) Troubleshooting 1.1 No output is produced plotmap, plotvoronoi, levelplot, and plotc use lattice functions. Therefore, in loops, functions, Sweave chunks, etc. the lattice object needs to be printed eplicitl b print (plotmap (object)) (R FAQ: Wh do lattice/trellis graphics not work?). The same holds for ggplot2 graphics. For suggestions how the lattice functions can be redefined so that the result is printed without eternal print command, see the file vignettes.defs. 11 Interactive Graphics hperspec offers basic interaction, spc.identif for spectra plots, and map.identif and map.sel.pol for maps. The first two identif points in spectra plots and map plots, respectivel. map.sel.pol selects the part of a hperspec object that lies inside the user defined polgon. 22

23 11.1 spc.identif: finding out wavelength, intensit and spectrum spc.identif allows to measure points in graphics produced b plotspc. It works correctl with reversed and cut wavelength aes. > spc.identif (plotspc (paracetamol, wl.range = c (6 ~ 18, 28 ~ 32), offset = 8)) The result is a data.frame with the indices of the spectra, the wavelength, and its intensit map.identif: finding a spectrum in a map plot map.identif returns the spectra indices of the clicked points. > map.identif (chondro) 11.3 map.sel.pol: selecting spectra inside a polgon in a map plot map.sel.pol returns a logical indicating which spectra are inside the polgon drawn b the user: > map.sel.pol (chondro) 11.4 Related functions provided b base graphics and lattice For base graphics (as produced b plotspc), locator ma be useful as well. It returns the clicked coordinates. Note that these are not transformed according to offset & Co. For lattice graphics, grid.locator ma be used instead. If it is not called in the panel function, a preceding call to trellis.focus is needed: > plot (laser, "mat") > trellis.focus () > grid.locator () identif (or panel.identif for lattice graphics) allows to identif points of the plot directl. Note that the returned indices correspond to the plotted object. References [1] Lemon J. Plotri: a package in the red light district of r. R-News, 6(4):8 12, 26. [2] Deepaan Sarkar and Feli Andrews. latticeetra: Etra Graphical Utilities Based on Lattice, 216. URL R package version [3] Rolf Turner. deldir: Delauna Triangulation and Dirichlet (Voronoi) Tessellation, 218. URL R package version.1-. [4] Fortran code b R. J. Renka. R functions b Albrecht Gebhardt. With contributions from Stephen Eglen <stephen@anc.ed.ac.uk>, Sergei Zuev, and Denis White. tripack: Triangulation of Irregularl Spaced Data, 216. URL R package version [] Daniel Adler, Duncan Murdoch, and others. rgl: 3D Visualization Using OpenGL, 218. URL R package version [6] Hadle Wickham. ggplot2: Elegant Graphics for Data Analsis. Springer-Verlag New York, 216. ISBN URL 23

24 Session Info [,1] ssname "Linu" release " generic" version "#37-Ubuntu SMP Mon Apr 18 18:33:37 UTC 216" nodename "cb-t61p" machine "86_64" login "unknown" user "cb" effective_user "cb" R version (218-3-) Platform: 86_64-pc-linu-gnu (64-bit) Running under: Ubuntu LTS Matri products: default BLAS: /usr/lib/openblas-base/libblas.so.3 LAPACK: /usr/lib/libopenblasp-r.2.18.so locale: [1] LC_CTYPE=de_DE.UTF-8 LC_NUMERIC=C LC_TIME=de_DE.UTF-8 [4] LC_COLLATE=C LC_MONETARY=de_DE.UTF-8 LC_MESSAGES=de_DE.UTF-8 [7] LC_PAPER=de_DE.UTF-8 LC_NAME=C LC_ADDRESS=C [1] LC_TELEPHONE=C LC_MEASUREMENT=de_DE.UTF-8 LC_IDENTIFICATION=C attached base packages: [1] tools grid stats graphics grdevices utils datasets methods base other attached packages: [1] latticeetra_.6-28 RColorBrewer_1.1-2 baseline_1.2-1 MASS_7.3- [] hperspec_ ggplot2_ lattice_.2-3 loaded via a namespace (and not attached): [1] reshape2_1.4.3 colorspace_1.3-2 testthat_2.. miniui_.1.1 [] htmltools_.3.6 XML_ rlang_.2. R.oo_1.22. [9] manipulatewidget_.9. pillar_1.2.2 later_.7.2 glue_1.2. [13] withr_2.1.2 R.utils_2.6. bindrcpp_.2.2 R.cache_.13. [17] bindr_.1.1 plr_1.8.4 stringr_1.3.1 munsell_.4.3 [21] gtable_.2. R.methodsS3_1.7.1 htmlwidgets_1.2 mvtnorm_1.-7 [2] labeling_.3 knitr_1.2 SparseM_1.77 httpuv_1.4.3 [29] crosstalk_1.. Rcpp_ table_1.8-2 scales_.. [33] promises_1..1 plotri_3.7 jsonlite_1. mime_. [37] deldir_.1- R.rsp_.42. digest_.6. stringi_1.2.2 [41] dplr_.7.4 shin_1.. magrittr_1. lazeval_.2.1 [4] tibble_1.4.2 pkgconfig_2..1 assertthat_.2. R6_2.2.2 [49] compiler_