Linearly Rescale Numeric Data

Performs a linear transformation of numeric data to a specified range. Each column of x is rescaled independently.

Usage

linScale(x, low = NULL, high = NULL, newLow = 0, newHigh = 1)

Arguments

x

a numeric vector, matrix or data frame.

low, high

optional numeric vectors specifying the lower and upper bounds of the original scale. If NULL, the column-wise minima and maxima of x are used.

newLow, newHigh

numeric vectors specifying the target range. Defaults to 0 and 1.

Value

an object of the same shape as x: a numeric vector for vector input, otherwise a numeric matrix with the same dimensions, where each column is linearly rescaled to the interval [newLow, newHigh].

Details

The transformation is defined as: $$ x_{scaled} = \frac{x - low}{high - low} \cdot (newHigh - newLow) + newLow $$

Constant columns (where high == low) are mapped to newLow.

If low and high are supplied, values of x outside [low, high] are extrapolated linearly and are not clipped to the target range.

See also

scale(), DescToolsX::scaleX

Other math.transform: logit(), percentRank(), rankX(), winsorize()

Examples

x <- matrix(1:10, ncol = 2)

# default scaling to [0,1]
linScale(x)
#>      [,1] [,2]
#> [1,] 0.00 0.00
#> [2,] 0.25 0.25
#> [3,] 0.50 0.50
#> [4,] 0.75 0.75
#> [5,] 1.00 1.00

# custom range
linScale(x, newLow = -1, newHigh = 1)
#>      [,1] [,2]
#> [1,] -1.0 -1.0
#> [2,] -0.5 -0.5
#> [3,]  0.0  0.0
#> [4,]  0.5  0.5
#> [5,]  1.0  1.0

# using predefined bounds
linScale(x, low = 1, high = 10)
#>           [,1]      [,2]
#> [1,] 0.0000000 0.5555556
#> [2,] 0.1111111 0.6666667
#> [3,] 0.2222222 0.7777778
#> [4,] 0.3333333 0.8888889
#> [5,] 0.4444444 1.0000000