Weighted Loss Function

weighted.loss() computes various loss metrics (e.g., RMSE, MAE) between two numeric vectors, or for the deviations from the weighted mean of a numeric vector.

Usage

weighted.loss(x, y = NULL, w = NULL, na.rm = FALSE, method = "rmse")

Arguments

x

a numeric vector.

y

an optional numeric vector. If NULL, x is compared against its weighted mean.

w

a numeric vector of sample weights for each value in x.

na.rm

logical. If TRUE, any NA and NaNs are removed from all input vectors before the calculation.

method

the loss measure. One of "mse" (mean square error), "rmse" (root mean square error), "mae" (mean absolute error), or "medae" (median absolute error).

Value

weighted.loss() returns a single numeric value.

Examples

# Calculate loss metrics between x and y with weights
weighted.loss(x = c(0, 10), y = c(0, 0), w = c(99, 1), method = "rmse")
#> [1] 1
weighted.loss(x = c(0, 10), y = c(0, 0), w = c(99, 1), method = "mae")
#> [1] 0.1
weighted.loss(x = c(0, 10), y = c(0, 0), w = c(99, 1), method = "medae")
#> [1] 0

# Verify uninterpreted variation ratio of a fitted MID model without weights
mid <- interpret(dist ~ speed, cars)
#> 'model' not passed: response variable in 'data' is used
RSS <- weighted.loss(cars$dist, predict(mid, cars), method = "mse")
TSS <- weighted.loss(cars$dist, method = "mse")
RSS / TSS
#> [1] 0.2263636
mid$ratio
#> [1] 0.2263636

# Verify uninterpreted variation ratio of a fitted MID model with weights
w <- 1:nrow(cars)
mid <- interpret(dist ~ speed, cars, weights = w)
#> 'model' not passed: response variable in 'data' is used
RSS <- weighted.loss(cars$dist, predict(mid, cars), w = w, method = "mse")
TSS <- weighted.loss(cars$dist, w = w, method = "mse")
RSS / TSS
#> [1] 0.2864814
mid$ratio
#> [1] 0.2864814