Computes the mean absolute error (MAE) between predicted
and reference values.
Usage
mae(x, ...)
# S3 method for class 'lm'
mae(x, ...)
# Default S3 method
mae(x, ref, na.rm = FALSE, ...)
Arguments
- x
an object. Methods are available for numeric vectors and
model objects (e.g. lm).
- ...
additional arguments passed to methods
- ref
numeric vector of reference (true) values
- na.rm
logical; whether to remove missing values
Value
a numeric scalar containing the MAE
Details
The mean absolute error is defined as:
$$
\frac{1}{n} \sum |ref - x|
$$
See also
mean(), abs()
Other model.metrics:
auc(),
averagePrecision(),
brierScore(),
logLoss(),
mape(),
mse(),
nmae(),
nmse(),
rmse(),
smape()
Examples
x <- c(2.5, 3.0, 2.8)
ref <- c(3.0, 2.5, 3.0)
mae(x, ref)
#> [1] 0.4
# with linear model
fit <- lm(mpg ~ hp, data = mtcars)
mae(fit)
#> [1] 2.907452
