Title:	High Dimensional Analysis in Linked Spaces
Version:	0.2.4
Description:	A 'shiny' GUI that performs high dimensional cluster analysis. This tool performs data preparation, clustering and visualisation within a dynamic GUI. With interactive methods allowing the user to change settings all without having to to leave the GUI. An earlier version of this package was described in Laa and Valencia (2022) <doi:10.1140/epjp/s13360-021-02310-1>.
License:	MIT + file LICENSE
Encoding:	UTF-8
RoxygenNote:	7.3.2
Depends:	R (≥ 3.5)
Imports:	tourr, stats, tibble, ggplot2, RColorBrewer, dplyr, dendextend, fpc, shiny, shinythemes, DT, tidyr, tidyselect, magrittr, detourr (≥ 0.2.0), crosstalk, Rtsne, plotly, alphahull, uwot, shinyFeedback, ggpcp, rlang, viridis
LazyData:	true
Suggests:	knitr, readr, rmarkdown, testthat (≥ 3.0.0), VIM
VignetteBuilder:	knitr
Config/testthat/edition:	3
URL:	https://gabrielmccoy.github.io/pandemonium/
NeedsCompilation:	no
Packaged:	2025-10-29 06:54:39 UTC; gabrielmccoy
Author:	Gabriel McCoy [aut, cre], Ursula Laa [aut], German Valencia [aut]
Maintainer:	Gabriel McCoy <gabe.mccoy02@gmail.com>
Repository:	CRAN
Date/Publication:	2025-11-03 08:30:02 UTC

Pipe operator

Description

See magrittr::%>% for details.

Usage

lhs %>% rhs

Value

No return value, called for side effects

Bike sharing data with model information

Description

The dataset contains daily counts of bikes rented with corresponding weather and seasonal information. The data is provided by Hadi Fanaee-T and available from https://doi.org/10.24432/C5W894. Additionally, model information from a single hidden layer neural network with eight nodes in the hidden layer has been added: the values of the activations for all observations (variables A1 to A8 in the cluster space) and the model prediction (pred) and residual (res) in the other variables.

Usage

Bikes

Format

a list of 4 dataframes

df

dataframe 731 obs of 18 variables containing the entire bikes data set

space1

dataframe 731 obs of 8 variables (cluster space)

space2

dataframe 731 obs of 6 variables (linked space, predictors used in the model)

other

dataframe 731 obs of 4 variables (other variables, including observed and predicted counts)

Generating layout for the graphical interface.

Description

(modified)

Usage

UI()

Value

shiny ui

Bin points based on chi2

Description

Map to values of sigma and compute equidistant binning in sigma.

Usage

chi2bins(chivals, ndf, k)

Arguments

Value

bin assignment for each point

Chi-squared scores function

Description

Can be used as getScores input in pandemonium. Returns chi-squared values as the score and sigma bins as the bins.

Usage

chi2score(space1, covinv, exp, ...)

Arguments

Value

named list containing scores for use in pandemonium

Examples

chi2score(Bikes$space1,solve(cov(Bikes$space1)),
            data.frame(value = colMeans(Bikes$space1)))

function for assigning colouring, palette and labels

Description

function for assigning colouring, palette and labels

Usage

colourHelper(choice, rv)

Arguments

Value

list containing colour assignment, palette and labels for use in plotting

Compute chi2 value for all points

Description

Compute chi2 value for all points

Usage

computeChi2(pred, covInv, exp)

Arguments

Value

vector with chi2 values

Compute sigma

Description

Map chi2 to sigma, with cutoff (overflow) at 5 sigma

Usage

computeSigma(chivals, ndf)

Arguments

Value

vector with sigma values

function for labeling cluster statistics on statistics page of pandemonium GUI

Description

function for labeling cluster statistics on statistics page of pandemonium GUI

Usage

cstat_labeller()

Compute cluster information

Description

The returned tibble contains the id of the cluster benchmark, the cluster radius and diameter, and group number for each cluster.

Usage

getBenchmarkInformation(dmat, groups)

Arguments

Value

data frame with cluster information

Examples

dists <- getDists(Bikes$space1,"euclidean")
fit <- stats::hclust(dists, "ward.D2")
groups <- stats::cutree(fit, k = 4)
getBenchmarkInformation(as.matrix(dists), groups)

Compute cluster distance summaries

Description

The returned tibble contains the id of the cluster pairs, with benchmark distance (d1), minimum (d2) and maximum (d3) distances between any points in the two clusters.

Usage

getClusterDists(dmat, groups, benchmarks)

Arguments

Value

data frame with distance information

Examples

dists <- getDists(Bikes$space1,"euclidean")
fit <- stats::hclust(dists, "ward.D2")
groups <- stats::cutree(fit, k = 4)
bm <- getBenchmarkInformation(as.matrix(dists), groups)
getClusterDists(as.matrix(dists), groups, bm)

Compute cluster statistics

Description

For number of clusters k between two and kmax, evaluate cluster statistics collected in output tibble.

Usage

getClusterStats(dist, fit, chivals, kmax = 10)

Arguments

Value

data frame with cluster statistics

Compute distances between all points

Description

Compute distances between all points

Usage

getDists(coord, metric, user_dist = NULL)

Arguments

Value

distances between all points

Examples

getDists(Bikes$space1[1:5,],"euclidean")
getDists(Bikes$space1[1:5,],"maximum")

Generate a specified plot outside the GUI

Description

An interface to generate a specific graph seen when using the GUI. Settings include: metric, linkage, k, plotType, for details see the vignette on using this function.

Usage

makePlots(
  space1,
  settings,
  cov = NULL,
  covInv = NULL,
  exp = NULL,
  space2 = NULL,
  space2.cov = NULL,
  space2.covInv,
  space2.exp = NULL,
  user_dist = NULL,
  getCoordsSpace1 = normCoords,
  getCoordsSpace2 = normCoords,
  getScore = NULL
)

Arguments

Value

ggplot, plotly or detourr plot depending on settings$plotType

Examples

makePlots(space1 = Bikes$space1,
  settings = list(
      plotType = "WC", x="hum", y="temp", k=4, metric="euclidean",
      linkage="ward.D2", WCa=0.5, showalpha=TRUE),cov = cov(Bikes$space1),
      space2 = Bikes$space2, getScore = outsidescore(Bikes$other$res,"Residual"))

makePlots(space1 = Bikes$space1,
  settings = list(
      plotType = "tour", k=4, metric="euclidean", linkage="ward.D2",
      tourspace="space1", colouring="clustering", out_dim=2, tour_path="grand",
      display="scatter",radial_start=NULL, radial_var=NULL, slice_width=NULL, seed = 2025),
      cov = cov(Bikes$space1), space2 = Bikes$space2,
      getScore = outsidescore(Bikes$other$res,"Residual"))

Scaled coordinates

Description

Using scale to center and scale the coordinates.

Usage

normCoords(df, ...)

Arguments

Value

matrix with coordinate representation of all points

Examples

head(normCoords(Bikes$space2))

Using externally computed score values

Description

Can be used as getScores input in pandemonium, to use score values that are computed externally. Returns scores values as the score, and bins computed as below, between or above the first and third quartile.

Usage

outsidescore(scores, scoreName = NULL)

Arguments

Value

named list containing scores for use in pandemonium

Examples

pandemonium(df = Bikes$space1, space2 = Bikes$space2,
              getScore = outsidescore(Bikes$other$res,"Residual"))

Shiny app for exploring clustering solutions

Description

Opening the GUI to cluster the data points based on values in space2. Coordinates and distances are computed on the fly, or can be entered in the function call.

Usage

pandemonium(
  df,
  cov = NULL,
  is.inv = FALSE,
  exp = NULL,
  space2 = NULL,
  space2.cov = NULL,
  space2.exp = NULL,
  group = NULL,
  label = NULL,
  user_dist = NULL,
  dimReduction = list(tSNE = tSNE, umap = umap),
  getCoords = list(normal = normCoords),
  getScore = NULL
)

Arguments

Value

No return value, called to initiate 'shiny' app

Plot chi2

Description

Parameter values no longer need to be on a regular grid pattern for this plot.(modified)

Usage

plotChi2(wc, chi2, x, y, scoreName = NULL, cond = NULL)

Arguments

Value

ggplot

Plot selected cluster statistics

Description

Plot selected cluster statistics

Usage

plotCstat(dist, fit, chivals, stat, kmax = 8)

Arguments

Value

ggplot

Plot dimension reduction plot

Description

Plot dimension reduction plot

Usage

plotDimRed(
  coord1,
  coord2,
  d_mat1,
  d_mat2,
  data,
  colouring,
  dimReduction,
  algorithm,
  group,
  score,
  user_group,
  pch,
  seed = NULL
)

Arguments

Value

plotly plot

Plot heatmap with dendrogram

Description

Plot heatmap with dendrogram

Usage

plotHeatmap(dat, fit, k, pal)

Arguments

Value

plot

Make coordinate plot

Description

Parameter values no longer need to be on a regular grid pattern for this plot.(modified)

Usage

plotObs(coord, x, y, wc, obs, cond = NULL)

Arguments

Value

ggplot

Make parallel coordinate plot

Description

Make parallel coordinate plot

Usage

plotPC(
  coord,
  groups,
  benchmarkIds,
  filt,
  c = TRUE,
  s = TRUE,
  a = 0.2,
  pal = NULL
)

Arguments

Value

ggplot

Plot sigma bins in parameter space

Description

Parameter values no longer need to be on a regular grid pattern for this plot.(modified)

Usage

plotSigBin(
  wc,
  interest,
  bmID,
  sigmabins,
  x,
  y,
  binName,
  cond = NULL,
  colourSet = "Set2"
)

Arguments

Value

ggplot

Show clusters in parameter space

Description

Parameter values no longer need to be on a regular grid pattern for this plot.(modified)

Usage

plotWC(
  wc,
  x,
  y,
  interest,
  bmID,
  col,
  cond = NULL,
  groups = NULL,
  pal = NULL,
  a = 0.2,
  showalpha = TRUE
)

Arguments

Value

ggplot

Chi-Squared Loss Function Coordinates

Description

Computes coordinate values by comparing observed values to the reference, using the covariance matrix as when computing the chi-squared loss.

Usage

pullCoords(df, covInv, exp, ...)

Arguments

Value

matrix with coordinate representation of all points

Examples

head(pullCoords(Bikes$space2,solve(cov(Bikes$space2)),
            data.frame(value = colMeans(Bikes$space2))))

Generic Loss Function Coordinates

Description

Coordinates are computed as centered by the reference value and scaled with the standard deviation. Uses the i,ith entry of the covariance matrix as the standard deviation of the ith variable.

Usage

pullCoordsNoCov(df, cov, exp, ...)

Arguments

Value

matrix with coordinate representation of all points

Examples

head(pullCoordsNoCov(Bikes$space2,cov(Bikes$space2),
                data.frame(value = colMeans(Bikes$space2))))

Raw coordinates

Description

Returns the input data frame. This is used when other coordinate computations fail. In general, scaling of the inputs is recommended before clustering.

Usage

rawCoords(df, ...)

Arguments

Details

Externally calculated coordinates can be used through userCoords or as input data with rawCoords used as the coordinate function. The use of userCoords over rawCoords is in the treatment of input data. As pandemonium displays the input data in many plots the use of coordinates as input data will result in these plots being less meaningful for interpretation. Use userCoords where coordinates are necessary to calculate distances but interpretation from plots of clustering space is necessary.

Value

matrix with coordinate representation of all points

Examples

head(rawCoords(Bikes$space2))

t-Distributed Stochastic Neighbor Embedding

Description

Computes non-linear dimension reduction with Rtsne and default parameters.

Usage

tSNE(dist, ...)

Arguments

Value

list containing a n x 2 matrix of reduced dimension data in Y

Examples

head(tSNE(getDists(Bikes$space1,"euclidean"))$Y)

function to make tours

Description

function to make tours

Usage

tourMaker(
  coord1,
  coord2,
  group,
  score,
  user_group,
  tourspace,
  colouring,
  out_dim,
  tour_path,
  display,
  radial_start = NULL,
  radial_var = NULL,
  slice_width = NULL,
  seed = NULL
)

Arguments

Value

detour

Uniform Manifold Approximation and Projection Embedding

Description

Computes non-linear dimension reduction with uwot and default parameters.

Usage

umap(dist, ...)

Arguments

Value

list containing a 2 x n matrix of reduced dimension data

Examples

head(umap(getDists(Bikes$space1,"euclidean"))$Y)

User defined coordinate function

Description

Allows the use of externally calculated coordinates in the app. Can only be used when variables are not reassigned between the two spaces.

Usage

userCoords(user_coords)

Arguments

Details

Externally calculated coordinates can be used through userCoords or as input data with rawCoords used as the coordinate function. The use of userCoords over rawCoords is in the treatment of input data. As pandemonium displays the input data in many plots the use of coordinates as input data will result in these plots being less meaningful for interpretation. Use userCoords where coordinates are necessary to calculate distances but interpretation from plots of clustering space is necessary.

Value

function that returns the user defined coordinates user_coords

Examples

pandemonium(df = Bikes$space1, space2 = Bikes$space2,
              coords = list(normalised = normCoords, space2 = userCoords(Bikes$space2)))

Write coordinates and cluster assignment to a CSV file

Description

For working with the results outside the app. Settings used: metric, linkage, k

Usage

writeResults(
  space1,
  cov = NULL,
  covInv = NULL,
  exp = NULL,
  space2,
  space2.cov = NULL,
  space2.covInv = NULL,
  space2.exp = NULL,
  settings,
  filename,
  user_dist = NULL,
  getCoords.space1 = normCoords,
  getCoords.space2 = rawCoords
)

Arguments

Value

No return value, called for writing file

Examples

file<-tempfile()
writeResults(space1 = Bikes$space1, space2 = Bikes$space2,
settings = list(metric="euclidean",linkage="ward.D2",k=4), filename = file)
file.remove(file)