rscale: Robustly scale a numeric vector.
In harrysouthworth/conclus: Consensus Clustering

Robustly scale a numeric vector.

1	rscale(x, center = TRUE, scale = TRUE, na.action = na.omit)

`x`	A numeric vector.
`center`	Logical, whether to center using the median. Defauts to `center = TRUE`.
`scale`	Logical, whether to scale using the mad. Defaults to `scale = TRUE`.
`na.action`	What to do when there are missing values. Defaults to `na.action=na.omit` and missing values are returned in the same places as in `x`. The alternative is `na.action=na.fail` which will cause the funtion to fail if there are missing values.

The vector will be centered by its median and then scaled by its median absolute deviation. There may be better (more efficient) ways of doing this (say by using M-estimators of location and scale), but for most cases, this approach ought to be fine.

harrysouthworth/conclus documentation built on May 24, 2019, 4:05 a.m.