Autocorrelation and distribution tests

Evaluate a Kolmogorov-Smirnov test on the residuals of a solarModel model object against the estimated Gaussian mixture distribution and a Breush-pagan or Box-pierce test on the residuals.

solarModel_tests(
  model,
  lags = c(7),
  method = "bg",
  type = "full",
  ci = 0.05,
  min_quantile = 0.025,
  max_quantile = 0.985
)

Arguments

model

An object of the class solarModel

lags

Numeric vector. Lags on which perform the autocorrelation tests.

method

character(1), method of test performed. Can be:

"bg"

for Breush-Godfrey test;

"bp"

for Box-pierce test;

"lb"

for Ljung-Box test.

type

character(1), type of test. Can be:

"train"

the test is performed only on train data;

"test"

the test is performed only on test data;

"full"

the test is performed only on the full data.

ci

p.value for rejection.

min_quantile

minimum quantile for the grid of values.

max_quantile

maximum quantile for the grid of values.

Note

Version 1.0.0.

Examples

model = solarModel$new(spec)
model$fit()
#> No outliers!
solarModel_tests(model, train_data = TRUE)
#> Error in solarModel_tests(model, train_data = TRUE): unused argument (train_data = TRUE)