pinball function

Pinball function

The function returns the value from the pinball function for the specified level and the type of loss

pinball(holdout, forecast, level, loss = 1, na.rm = TRUE)

Arguments

• holdout: The vector or matrix of the holdout values.
• forecast: The forecast of a distribution (e.g. quantile or expectile). It should be the same length as the holdout.
• level: The level associated with the forecast (e.g. level of quantile).
• loss: The type of loss to use. The number which corresponds to L1, L2 etc. L1 implies the loss for quantiles, while L2 is for the expectile.
• na.rm: Logical, defining whether to remove the NAs from the provided data or not.

Returns

The function returns the scalar value.

Examples

# An example with mtcars data
ourModel <- alm(mpg~., mtcars[1:30,], distribution="dnorm")

# Produce predictions with the interval
ourForecast <- predict(ourModel, mtcars[-c(1:30),], interval="p")

# Pinball with the L1 (quantile value)
pinball(mtcars$mpg[-c(1:30)],ourForecast$upper,level=0.975,loss=1)
pinball(mtcars$mpg[-c(1:30)],ourForecast$lower,level=0.025,loss=1)

# Pinball with the L2 (expectile value)
pinball(mtcars$mpg[-c(1:30)],ourForecast$upper,level=0.975,loss=2)
pinball(mtcars$mpg[-c(1:30)],ourForecast$lower,level=0.025,loss=2)

Author(s)

Ivan Svetunkov, ivan@svetunkov.com