rescale: Rescale a numeric vector
In danielmarcelino/SciencesPo: A Tool Set for Analyzing Political Behavior Data

Description Usage Arguments Details Value Examples

rescale returns a rescaled numeric vector.

1 2	rescale(x, method = c("normal", "minmax"), mean = 0, sd = 1, min = 0, max = 1)

`x`	A numeric vector.
`method`	A character string indicating the desired method of rescaling with `"normal"` being the default. This must be (an abbreviation of) one of the strings `"normal"` or `"minmax"`. See Details for more information.
`mean`	The desired mean value for normal-style scaling. Used only when `method="normal"`.
`sd`	The desired standard deviation value for normal-style scaling. Used only when `method="normal"`.
`min`	The desired minimum value for min/max-style scaling. Used only when `method="minmax"`.
`max`	The desired maximum value for min/max-style scaling. Used only when `method="minmax"`.

rescale returns a rescaled version of the numeric vector x. NA values in x are ignored during the rescaling process but are preserved in the output.

rescale is designed to be readable from the function call. For example:

rescale(x, "normal", mean=0, sd=1) can be read as "Rescale x using a normal-style transformation with mean 0 and standard deviation 1".
rescale(x, "minmax", min=0, max=1) can be read as "Rescale x using a min/max-style transformation with min 0 and max 1".

The arguments mean, sd, min, and max are used based on method. rescale offers a couple of different options for method:

normal: The default option that rescales x using a normal-style transformation into a distribution with mean mean and standard deviation sd. The default values for this rescaling are mean=0 and sd=1.

The explicit formula for this transformation is:

(x - μ_x)/(σ_x) * σ + μ

where μ_x is the sample mean of x, σ_x is the sample standard deviation of x, μ is the desired mean, and σ is the desired standard deviation.
minmax: rescales x using a min/max-style transformation into a distribution with minimum min and maximum max. The default values for this rescaling are min=0 and max=1.

The explicit formula for this transformation is:

((x - min_x)/(max_x - min_x)) * (max - min) + min

where min_x is the sample minimum of x, max_x is the sample maximum of x, min is the desired minimum, and max is the desired maximum.

The output of rescale is a rescaled numeric vector with the same length as x.

set.seed(1337)
x <- rnorm(1e4)
summary(x)

#####
# Common use cases
#

# I want to rescale to a standard normal distribution!
x_normal <- rescale(x)
summary(x_normal)
mean(x_normal); sd(x_normal)

# I want to rescale to be between 0 and 1!
x_minmax <- rescale(x, "minmax")
summary(x_minmax)
min(x_minmax); max(x_minmax)

# I want to rescale to be between 300 and 850! (Weird but some credit scores do it!)
x_credit <- rescale(x, "minmax", min=300, max=850)
summary(x_credit)
min(x_credit); max(x_credit)