PIT_local {recalibratiNN}

Obtain local PIT-values from a model

Description

Return local PIT-values. Centroids for localization is obtained by k-means method from stats package. The vicinity of such centroids are selected though a aproximate k-nearst neighboors method from RANN package.

Usage

PIT_local(
  xcal,
  ycal,
  yhat,
  mse,
  clusters = 6,
  p_neighbours = 0.2,
  PIT = PIT_global
)

Arguments

xcal

features/covariates from calibration set

ycal

observations of calibration set

yhat

predicted outputs from the calibrations et

mse

Mean Squared Error of the model

clusters

Number of partitions to create for local calibration. Centroids calculated by k-means method. Default set to 6.

p_neighbours

Proportion of xcal to localize neighboors in the KNN method. Default is 0.2.

PIT

function to return the PIT-values. Default set to PIT_global() from this package.

Value

A tibble with five containing in the first column containing unique names for the partition, "y_cal", the second column containing the yhat the third the pit-values and the last the number of neighbors in each partition.

Examples

n <- 10000
split <- 0.8

mu <- function(x1){
10 + 5*x1^2
}

sigma_v <- function(x1){
 30*x1
}

x <- runif(n, 1, 10)
y <- rnorm(n, mu(x), sigma_v(x))

x_train <- x[1:(n*split)]
y_train <- y[1:(n*split)]

x_cal <- x[(n*split+1):n]
y_cal <- y[(n*split+1):n]

model <- lm(y_train ~ x_train)

y_hat <- predict(model, newdata=data.frame(x_train=x_cal))

MSE_cal <- mean((y_hat - y_cal)^2)

PIT_local(xcal = x_cal, ycal=y_cal, yhat=y_hat, mse=MSE_cal)