Help for package klustR

Title:

D3 Dynamic Cluster Visualizations

Description:

Used to create dynamic, interactive 'D3.js' based parallel coordinates and principal component plots in 'R'. The plots make visualizing k-means or other clusters simple and informative.

Version:

0.1.0

URL:

https://mckaymdavis.github.io/klustR/, https://github.com/McKayMDavis/klustR

BugReports:

https://github.com/McKayMDavis/klustR/issues

License:

GPL (≥ 3)

Depends:

R (≥ 3.6.0)

Imports:

htmlwidgets (≥ 0.3.2), jsonlite

Encoding:

UTF-8

LazyData:

true

RoxygenNote:

6.1.1

NeedsCompilation:

Packaged:

2019-06-18 18:01:08 UTC; mckaydavis

Author:

McKay Davis [aut, cre]

Maintainer:

McKay Davis <mckaymdavis@gmail.com>

Repository:

CRAN

Date/Publication:

2019-06-19 09:40:08 UTC

Shiny bindings for klustR widgets

Description

Output and render functions for using klustR widgets within Shiny applications and interactive Rmd documents.

Usage

pacoplotOutput(outputId, width = "100%", height = "400px")

renderpacoplot(expr, env = parent.frame(), quoted = FALSE)

pcplotOutput(outputId, width = "100%", height = "400px")

renderpcplot(expr, env = parent.frame(), quoted = FALSE)

Arguments

outputId

output variable to read from

width, height

Must be a valid CSS unit (like "100%", "400px", "auto") or a number, which will be coerced to a string and have "px" appended.

expr

An expression that generates a klustR graph.

env

The environment in which to evaluate expr.

quoted

Is expr a quoted expression (with quote())? This is useful if you want to save an expression in a variable.

Parallel Coordinates Plot for Clustering

Description

Creates an interactive parallel coordinates plot detailing each dimension and the cluster associated with each observation.

Usage

pacoplot(data, clusters, colorScheme = "schemeCategory10",
  width = NULL, height = NULL, labelSizes = NULL, lineSize = NULL,
  measures = NULL)

Arguments

data

A dataframe of numeric columns.

clusters

A named integer matrix of clusters where names are the row names of the above dataframe and integers are the integer value of the row's associated cluster. This can be obtained from a function such as stats::kmeans()$cluster .

colorScheme

The color scheme of the plot. May be a pre-configured D3 ordinal color scheme or a vector of html colors (hex or named) of the same length as the number of clusters.

width

The width of the plot window.

height

The height of the plot window.

labelSizes

A number or list of any combination of parameters shown that define the label sizes. list(yaxis = 12, yticks = 10, tooltip = 15)

lineSize

A number to adjust the size of the lines.

measures

A list of functions that is any combination of parameters shown that define the measurements for intervals and average lines displayed. Defaults to the options shown (median and 1st and 3rd quartile).
list(avg = median, upper = function(x){return(quantile(x, c(0.75)))}, lower = function(x){return(quantile(x, c(0.25)))})

Details

Hover over lines to display row label
Click on a line to fade out all lines except the associated cluster
Click on another line to bold this line as well
Clicking a second time on a line will fade it out

Examples


# Barebones
df <- state.x77
clus <- kmeans(df, 5)$cluster
pacoplot(data = df, clusters = clus)

# With options
df <- state.x77
clus <- kmeans(df, 5)$cluster
pacoplot(data = df, clusters = clus,
         colorScheme = c("red", "green", "orange", "blue", "yellow"),
         labelSizes = list(yaxis = 16, yticks = 12),
         measures = list(avg = mean))

Principal Component Plot for K-Means Clustering

Description

Reduces dimensionality to 2D using principal component analysis (PCA) and displays a dynamic visualization of two principal components (PC).

Usage

pcplot(data, clusters, barColor = "steelblue",
  colorScheme = "schemeCategory10", width = NULL, height = NULL,
  labelSizes = NULL, dotSize = NULL, pcGridlines = FALSE,
  barGridlines = FALSE)

Arguments

data

A dataframe of numeric columns. Scaled data is preferred as PCA does not work the same with non-scaled data.

clusters

barColor

The color to use for the bar-chart fill. May be any html color (hex or named).

colorScheme

The color scheme of the PCA plot. May be a pre-configured D3 ordinal color scheme or a vector of html colors (hex or named) of the same length as the number of clusters.

width

The width of the plot window.

height

The height of the plot window.

labelSizes

A number or list of any combination of parameters shown that define the label sizes.
list(yaxis = 12, yticks = 10, tooltip = 15)

dotSize

A number to adjust the size of the dots.

pcGridlines

TRUE FALSE Show grid-lines on the PC plots?

barGridlines

TRUE FALSE Show grid-lines on the bar-charts?

Details

Clicking on axis labels will display a bar-chart of PC contribution
Clicking on legend colors will fade out all points but the points in the cluster selected
Hover over points to see the label and point coordinates

Examples


# Barebones
scaled_df <- scale(state.x77)
clus <- kmeans(scaled_df, 5)$cluster
pcplot(data = scaled_df, clusters = clus)

# With Options
scaled_df <- scale(state.x77)
clus <- kmeans(scaled_df, 5)$cluster
pcplot(data = scaled_df, clusters = clus,
       barColor = "red",
       colorScheme = c("red", "green", "orange", "blue", "yellow"),
       labelSizes = list(yaxis = 20, yticks = 15, tooltip = 25),
       pcGridlines = TRUE, barGridlines = TRUE)