rgr: gx.symm.coords.plot – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

gx.symm.coords.plot

Display a X-Y Plot, Scatterplot, of Symmetric Balances

Description

Displays a scatterplot based on symmetric balances for two parts, v1 and v2, from a n by p matrix or data frame of compositional data, and their correlation coefficient by the selected method, default spearman. Optionally the display in Garrett et al. (2017) may be generated, in which case the correlation coefficients estimated by any valid R method, by default "spearman" are displayed above the plots. By default the parts are displayed, left plot, with log-scaling. To plot without log-scaling, set log = FALSE, in which case the correlation coefficient is estimated without a logarithmic transformation. If "pearson" coefficients are required the option of a log transformation for the data is provided.

Usage

gx.symm.coords.plot(x, v1, v2, log = TRUE, method = "spearman", example = " ", ... )

Arguments

`x`	`n` by `p` matrix or data frame of compositional data for which a scatterplot will be displayed for two parts.
`v1`	the column index for the part to be displayed on the x-axis.
`v2`	the column index for the part to be displayed on the y-axis.
`log`	to compute the non-symmetric balance coefficients with a logarithmic transformation, set `log = TRUE`.
`method`	the valid R method for computation of the correlation coefficient between the symmetric balances for `v1` and `v2`, the default is `"spearman"`.
`example`	to reproduce the display in Garrett et al. (2017) set `example = "Explore"` or `"explore"`.
`...`	further arguments to be passed to methods concerning the plot.

Note

For compositional data analysis all the data must be in the same measurement units.

The "spearman" coefficient is preferred for EDA as any systematic monotonic variation in the data is of interest, and may be worthy of further investigation. As "spearman" coefficients are based on ranks, any monotonic data transformation, e.g., logarithmic, has no impact on the results. This is not the case for "pearson" coefficients.

Author(s)

Robert G. Garrett

References

Garrett, R.G., Reimann, C., Hron, K., Kynclova, P. and Filzmoser, P., 2017. Finally, a correlation coefficient that tells the geochemical truth. Explore - Assoc. Applied Geochemists Newsletter, 176:1-10.

Reimann, C., Filzmoser, P., Hron, K., Kynclova, P. and Garrett, R.G., 2017. Correlation Analysis for Compositional (Environmental) Data. Science of the Total Environment, 607-608:965-971.

Examples

## Make test data available
data(nockolds)

## Display symmetric balances scatterplot for variable 1, Si, and 2, Al, for the nockolds data
gx.symm.coords.plot(nockolds, 1, 2)

## Display the example from Garrett et al. (2017)
gx.symm.coords.plot(nockolds, 1, 2, example = "explore")

## Clean-up
rm(nockolds)

rgr

Applied Geochemistry EDA

v1.1.15

GPL-2

Authors

Robert G. Garrett

Initial release

2018-03-05