Measure to compare true observed response with predicted response in regression tasks.
Arguments
- truth
(
numeric()
)
True (observed) values. Must have the same length asresponse
.- response
(
numeric()
)
Predicted response values. Must have the same length astruth
.- sample_weights
(
numeric()
)
Vector of non-negative and finite sample weights. Must have the same length astruth
. The vector gets automatically normalized to sum to one. Defaults to equal sample weights.- alpha
numeric(1)
The quantile to compute the pinball loss.- ...
(
any
)
Additional arguments. Currently ignored.
Details
The pinball loss for quantile regression is defined as $$ \text{Average Pinball Loss} = \frac{1}{n} \sum_{i=1}^{n} w_{i} \begin{cases} q \cdot (t_i - r_i) & \text{if } t_i \geq r_i \\ (1 - q) \cdot (r_i - t_i) & \text{if } t_i < r_i \end{cases} $$ where \(q\) is the quantile and \(w_i\) are normalized sample weights.
Meta Information
Type:
"regr"
Range: \((-\infty, \infty)\)
Minimize:
TRUE
Required prediction:
response