iplotCorr: Correlation heatmap

In COINr: Composite Indicator Construction and Analysis

Description

Plots an interactive heatmap of a correlation matrix. Currently this only works with the aggregated data set, i.e. you need to have aggregated the data first before using this.

Usage

iplotCorr(
  COIN,
  aglevs = NULL,
  insig = FALSE,
  levs = TRUE,
  grouprects = TRUE,
  flagcolours = TRUE,
  corthresh = NULL,
  showvals = TRUE,
  cortype = "pearson",
  useweights = NULL
)

Arguments

COIN	The COIN object
aglevs	A two length vector specifying which level to plot against which level. E.g. c(2,4) for COIN plots sub-pillars against sub-indexes. If NULL, plots everything against everything.
insig	Logical: if TRUE, all correlation values are plotted; if FALSE (default), does not plot insignificant correlations.
levs	Logical: if TRUE, plots lines showing the division between different levels. Only works if aglevs = NULL.
grouprects	Logical: if TRUE, plots rectangles showing aggregation groups.
flagcolours	If TRUE uses a discrete colour scale specified by corthresh. Otherwise uses a continuous colour map.
corthresh	A named list specifying the colour thresholds to use if flagcolours = TRUE. Entries should specify correlation thresholds and can specify any of clow, cmid and chi. Anything below clow will be coloured red. Anything between clow and cmid will be grey. Anything between cmid and chigh will be blue. Anything above chigh will be green. Default is list(clow = -0.4, cmid = 0.4, chigh = 0.85). You can specify a subset of these and the others will revert to defaults.
showvals	Logical: if TRUE, overlays correlation values on each square.
cortype	The type of correlation: either "pearson" (default), "spearman" or "kendall". See stats::cor.
useweights	An optional list of weights to use (this is used mainly in the rew8r() app).

Details

This is a slightly involved wrapper for plotly. It allows plotting any level against any other, and outputs correlation heat maps as HTML widgets. It has some flexibility regarding grouping of indicators, colouring, treatment of insignificant correlations, and the correlation type. Explore the arguments and see.

Value

A plotly correlation map (figure).

See Also

• plotCorr() Static correlation heat maps

• rew8r() Interactive app for adjusting weights and seeing effects on correlations

• getCorr() Get correlations between indicators/levels

Examples

# build ASEM COIN up to aggregation
ASEM <- build_ASEM()
# correlation heatmap of pillars against sub-indexes
iplotCorr(ASEM, aglevs = c(2,3))

COINr documentation built on Nov. 30, 2021, 9:06 a.m.