Compute the weights for robust loess, for all robustness levels <= the
robustness parameter. This computation is embarrassingly parallel, so if a
TaskPool is provided it will be parallelized.
This function must be called before calling predict() with a robustness
value > 0. computeRobustWeights() must be called with a robustness level
>= the robustness level predict() is to be called with. This is not
handled implicitly because computeRobustWeights() is very computationally
intensive and because it modifies the state of the Loess1D object, while
predict() is const. Forcing computeRobustWeights() to be called explicitly
allows multiple instances of predict() to be evaluated in parallel.
Compute the weights for robust loess, for all robustness levels <= the robustness parameter. This computation is embarrassingly parallel, so if a TaskPool is provided it will be parallelized.
This function must be called before calling predict() with a robustness value > 0. computeRobustWeights() must be called with a robustness level >= the robustness level predict() is to be called with. This is not handled implicitly because computeRobustWeights() is very computationally intensive and because it modifies the state of the Loess1D object, while predict() is const. Forcing computeRobustWeights() to be called explicitly allows multiple instances of predict() to be evaluated in parallel.