Nothing
tests/testthat/test-iterative_estimators2.R
In robust2sls: Outlier Robust Two-Stage Least Squares Inference and Testing
# new test file to find unit tests more easily
# testing iis_init in iterated algorithm
test_that("outlier_detection() with iis_init() works, fixed iterations", {
skip_on_cran()
# base setup
set.seed(10)
p <- generate_param(1, 1, 1, beta = c(2, 4), sigma = 1,
mean_z = 0, cov_z = matrix(1),
Sigma2_half = matrix(1), Omega2 = matrix(3/4),
Pi = t(matrix(c(1, 0, 0, 1), nrow = 2)))
d <- generate_data(parameters = p, n = 50)$data
formula <- y ~ -1+x1+x2 | -1+x1+z2
gamma <- 0.05
# iis arguments
arglist1 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = NULL)
model1 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 0,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# same setup as previous testing (iis_init and selection_iis)
# should have retained indicator for observation 19 in m = 0
## check overall structure
expect_identical(class(model1), "robust2sls")
expect_length(model1, 6)
expect_named(model1, c("cons", "model", "res", "stdres", "sel", "type"))
expect_type(model1, "list")
iis1 <- ivgets::ivisat(formula = formula, data = d, iis = TRUE,
t.pval = gamma, wald.pval = arglist1$t.pval,
do.pet = arglist1$do.pet, overid = arglist1$overid,
weak = arglist1$weak, print.searchinfo = FALSE)
res <- iis1$final$residuals
stdres <- res / (iis1$final$sigma * sqrt(iis1$final$df.residual / iis1$final$nobs))
sel <- rep(TRUE, times = NROW(d))
sel[19] <- FALSE
type <- rep(1L, times = NROW(d))
type[19] <- 0L
fullmodel <- ivreg::ivreg(formula = formula, data = d)
fullmodel$call <- NULL # won't coincide exactly
model1$model$m0$call <- NULL
names(res) <- names(stdres) <- names(sel) <- names(type) <- as.character(1:50)
expect_identical(model1$res$m0, res)
expect_identical(model1$stdres$m0, stdres)
expect_identical(model1$sel$m0, sel)
expect_identical(model1$type$m0, type)
expect_identical(model1$model$m0, fullmodel)
# iterate fixed number of times
model2 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = 1,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# will have to re-estimate model based on selection (i.e. leaving out obs 19)
# do manually to check whether correct
iterd <- d$selection <- model1$sel$m0
iterm1 <- ivreg::ivreg(formula = formula, data = d, subset = selection)
expect_identical(iterm1$nobs, 49L) # quick check that only 49 obs used in estim
# check residuals, whether any outside cutoff
res <- d[, "y"] - stats::predict(iterm1, newdata = d)
expect_identical(length(res), 50L) # residual for all, even obs 19 that was not used in estimation
psi <- 1 - gamma
cutoff <- stats::qnorm(p = 1-(gamma/2), mean = 0, sd = 1)
bias_corr <- 1/(((1-gamma)-2*cutoff*stats::dnorm(cutoff,mean=0,sd=1))/
(1-gamma))
sigma <- iterm1$sigma * sqrt(iterm1$df.residual / iterm1$nobs) * sqrt(bias_corr)
stdres <- res / sigma
sel <- (abs(stdres) <= cutoff)
type <- ifelse(sel == TRUE, 1L, 0L)
names(res) <- names(stdres) <- names(sel) <- names(type) <- as.character(1:50)
expect_identical(model2$res$m1, res)
expect_identical(model2$stdres$m1, stdres)
expect_identical(model2$sel$m1, sel)
expect_identical(model2$type$m1, type)
# iterate more often
model3 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = 10,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
expect_length(model3$res, 11)
# snapshot outputs
expect_snapshot_output(model1)
expect_snapshot_output(model2)
expect_snapshot_output(model3)
})
test_that("outlier_detection() with iis_init() works, convergence", {
skip_on_cran()
# base setup
set.seed(10)
p <- generate_param(1, 1, 1, beta = c(2, 4), sigma = 1,
mean_z = 0, cov_z = matrix(1),
Sigma2_half = matrix(1), Omega2 = matrix(3/4),
Pi = t(matrix(c(1, 0, 0, 1), nrow = 2)))
d <- generate_data(parameters = p, n = 50)$data
formula <- y ~ -1+x1+x2 | -1+x1+z2
gamma <- 0.05
# iis arguments
arglist1 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = NULL)
# iterate until convergence, convergence_criterion 0
model1 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = "convergence",
convergence_criterion = 0, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# converges at iteration 2 (technically already at iteration 1 but IIS stores full sample model m0)
expect_length(model1$model, 3)
expect_length(model1$res, 3)
expect_length(model1$stdres, 3)
expect_length(model1$sel, 3)
expect_length(model1$type, 3)
expect_identical(model1$cons$convergence$converged, TRUE)
expect_identical(model1$cons$convergence$criterion, 0)
expect_identical(model1$cons$convergence$difference, 0)
expect_identical(model1$cons$iterations$setting, "convergence")
expect_identical(model1$cons$iterations$actual, 2)
# iterate until convergence, convergence_criterion 0.5
model2 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = "convergence",
convergence_criterion = 0.5, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# converges at iteration 1 (technically already at iteration 0 but IIS stores full sample model m0)
expect_length(model2$model, 2)
expect_length(model2$res, 2)
expect_length(model2$stdres, 2)
expect_length(model2$sel, 2)
expect_length(model2$type, 2)
expect_identical(model2$cons$convergence$converged, TRUE)
expect_identical(model2$cons$convergence$criterion, 0.5)
expect_identical(model2$cons$iterations$setting, "convergence")
expect_identical(model2$cons$iterations$actual, 1)
# check whether was correct to stop
difference <- sum((model2$model$m0$coefficients - model2$model$m1$coefficients)^2)
expect_identical(model2$cons$convergence$difference, difference)
# iterate 5 times but with convergence_criterion 0, so should stop at 2
model3 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = 0, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# converges at iteration 2 (technically already at iteration 1 but IIS stores full sample model m0)
expect_length(model3$model, 3)
expect_length(model3$res, 3)
expect_length(model3$stdres, 3)
expect_length(model3$sel, 3)
expect_length(model3$type, 3)
expect_identical(model3$cons$convergence$converged, TRUE)
expect_identical(model3$cons$convergence$criterion, 0)
expect_identical(model3$cons$convergence$difference, 0)
expect_identical(model3$cons$iterations$setting, 5)
expect_identical(model3$cons$iterations$actual, 2)
})
test_that("outlier_detection() works with iis_init(), tests turned on", {
skip_on_cran()
# base setup
set.seed(50)
p <- generate_param(dx1 = 1, dx2 = 1, dz2 = 2, intercept = TRUE,
beta = c(2, 4), sigma = 1, mean_z = matrix(c(0,0), 2, 1),
cov_z = matrix(c(1,0,0,1), 2, 2),
Sigma2_half = matrix(1), Omega2 = matrix(3/4),
Pi = t(matrix(c(1, 0, 0, 1, 0, 1), nrow = 2)))
d <- generate_data(parameters = p, n = 50)$data
formula <- y ~ -1+x1+x2 | -1+x1+z2+z3
gamma <- 0.05
# checked by stepping through function call that weak instrument test is actually applied
# iis arguments, no testing
arglist1 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = NULL)
model1 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# outliers: m0 = 5, m1 = 4, m2 = 2, m3 = 2, m4 = 2, m5 = 2
# add weak instrument test, should not pass
arglist2 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = 0.00000000001)
expect_warning(expect_error(outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist2),
"IIS final model is NULL. See warning."))
# add overid test, should not pass
arglist3 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = 0.99, weak = NULL)
expect_warning(expect_error(outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist3),
"IIS final model is NULL. See warning."))
# add both tests
arglist4 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = 0.99, weak = 0.01)
expect_warning(expect_error(outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist4),
"IIS final model is NULL. See warning."))
# try the built-in tests from gets
# normality test
arglist5 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = 0.05,
turbo = FALSE, overid = NULL, weak = NULL)
model5 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist5)
# differs from model1 b/c of testing
# one indicator retained in iter 0, and then no further outlier detected in iterations
# check this manually
manual5 <- ivgets::ivisat(formula = formula, data = d, iis = TRUE,
t.pval = gamma, normality.JarqueB = 0.05,
print.searchinfo = FALSE)
# only second block is selected over, first one probably fails (is issue in gets)
# one retained in second block (39)
# check that diagnostics are present, normality test has been used
expect_identical(rownames(manual5$selection$diagnostics),
c("Ljung-Box AR(1)", "Ljung-Box ARCH(1)", "Jarque-Bera"))
sel5 <- rep(TRUE, times = 50)
sel5[39] <- FALSE
names(sel5) <- as.character(1:50)
expect_identical(model5$sel$m0, sel5)
# PET test
arglist6 <- list(t.pval = 0.0000001, do.pet = TRUE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = NULL)
model6 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist6)
expect_snapshot_output(model1)
expect_snapshot_output(model5)
expect_snapshot_output(model6)
})
Try the robust2sls package in your browser
Any scripts or data that you put into this service are public.
robust2sls documentation built on Jan. 11, 2023, 5:13 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
# new test file to find unit tests more easily
# testing iis_init in iterated algorithm
test_that("outlier_detection() with iis_init() works, fixed iterations", {
skip_on_cran()
# base setup
set.seed(10)
p <- generate_param(1, 1, 1, beta = c(2, 4), sigma = 1,
mean_z = 0, cov_z = matrix(1),
Sigma2_half = matrix(1), Omega2 = matrix(3/4),
Pi = t(matrix(c(1, 0, 0, 1), nrow = 2)))
d <- generate_data(parameters = p, n = 50)$data
formula <- y ~ -1+x1+x2 | -1+x1+z2
gamma <- 0.05
# iis arguments
arglist1 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = NULL)
model1 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 0,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# same setup as previous testing (iis_init and selection_iis)
# should have retained indicator for observation 19 in m = 0
## check overall structure
expect_identical(class(model1), "robust2sls")
expect_length(model1, 6)
expect_named(model1, c("cons", "model", "res", "stdres", "sel", "type"))
expect_type(model1, "list")
iis1 <- ivgets::ivisat(formula = formula, data = d, iis = TRUE,
t.pval = gamma, wald.pval = arglist1$t.pval,
do.pet = arglist1$do.pet, overid = arglist1$overid,
weak = arglist1$weak, print.searchinfo = FALSE)
res <- iis1$final$residuals
stdres <- res / (iis1$final$sigma * sqrt(iis1$final$df.residual / iis1$final$nobs))
sel <- rep(TRUE, times = NROW(d))
sel[19] <- FALSE
type <- rep(1L, times = NROW(d))
type[19] <- 0L
fullmodel <- ivreg::ivreg(formula = formula, data = d)
fullmodel$call <- NULL # won't coincide exactly
model1$model$m0$call <- NULL
names(res) <- names(stdres) <- names(sel) <- names(type) <- as.character(1:50)
expect_identical(model1$res$m0, res)
expect_identical(model1$stdres$m0, stdres)
expect_identical(model1$sel$m0, sel)
expect_identical(model1$type$m0, type)
expect_identical(model1$model$m0, fullmodel)
# iterate fixed number of times
model2 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = 1,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# will have to re-estimate model based on selection (i.e. leaving out obs 19)
# do manually to check whether correct
iterd <- d$selection <- model1$sel$m0
iterm1 <- ivreg::ivreg(formula = formula, data = d, subset = selection)
expect_identical(iterm1$nobs, 49L) # quick check that only 49 obs used in estim
# check residuals, whether any outside cutoff
res <- d[, "y"] - stats::predict(iterm1, newdata = d)
expect_identical(length(res), 50L) # residual for all, even obs 19 that was not used in estimation
psi <- 1 - gamma
cutoff <- stats::qnorm(p = 1-(gamma/2), mean = 0, sd = 1)
bias_corr <- 1/(((1-gamma)-2*cutoff*stats::dnorm(cutoff,mean=0,sd=1))/
(1-gamma))
sigma <- iterm1$sigma * sqrt(iterm1$df.residual / iterm1$nobs) * sqrt(bias_corr)
stdres <- res / sigma
sel <- (abs(stdres) <= cutoff)
type <- ifelse(sel == TRUE, 1L, 0L)
names(res) <- names(stdres) <- names(sel) <- names(type) <- as.character(1:50)
expect_identical(model2$res$m1, res)
expect_identical(model2$stdres$m1, stdres)
expect_identical(model2$sel$m1, sel)
expect_identical(model2$type$m1, type)
# iterate more often
model3 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = 10,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
expect_length(model3$res, 11)
# snapshot outputs
expect_snapshot_output(model1)
expect_snapshot_output(model2)
expect_snapshot_output(model3)
})
test_that("outlier_detection() with iis_init() works, convergence", {
skip_on_cran()
# base setup
set.seed(10)
p <- generate_param(1, 1, 1, beta = c(2, 4), sigma = 1,
mean_z = 0, cov_z = matrix(1),
Sigma2_half = matrix(1), Omega2 = matrix(3/4),
Pi = t(matrix(c(1, 0, 0, 1), nrow = 2)))
d <- generate_data(parameters = p, n = 50)$data
formula <- y ~ -1+x1+x2 | -1+x1+z2
gamma <- 0.05
# iis arguments
arglist1 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = NULL)
# iterate until convergence, convergence_criterion 0
model1 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = "convergence",
convergence_criterion = 0, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# converges at iteration 2 (technically already at iteration 1 but IIS stores full sample model m0)
expect_length(model1$model, 3)
expect_length(model1$res, 3)
expect_length(model1$stdres, 3)
expect_length(model1$sel, 3)
expect_length(model1$type, 3)
expect_identical(model1$cons$convergence$converged, TRUE)
expect_identical(model1$cons$convergence$criterion, 0)
expect_identical(model1$cons$convergence$difference, 0)
expect_identical(model1$cons$iterations$setting, "convergence")
expect_identical(model1$cons$iterations$actual, 2)
# iterate until convergence, convergence_criterion 0.5
model2 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = "convergence",
convergence_criterion = 0.5, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# converges at iteration 1 (technically already at iteration 0 but IIS stores full sample model m0)
expect_length(model2$model, 2)
expect_length(model2$res, 2)
expect_length(model2$stdres, 2)
expect_length(model2$sel, 2)
expect_length(model2$type, 2)
expect_identical(model2$cons$convergence$converged, TRUE)
expect_identical(model2$cons$convergence$criterion, 0.5)
expect_identical(model2$cons$iterations$setting, "convergence")
expect_identical(model2$cons$iterations$actual, 1)
# check whether was correct to stop
difference <- sum((model2$model$m0$coefficients - model2$model$m1$coefficients)^2)
expect_identical(model2$cons$convergence$difference, difference)
# iterate 5 times but with convergence_criterion 0, so should stop at 2
model3 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = 0.05, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = 0, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# converges at iteration 2 (technically already at iteration 1 but IIS stores full sample model m0)
expect_length(model3$model, 3)
expect_length(model3$res, 3)
expect_length(model3$stdres, 3)
expect_length(model3$sel, 3)
expect_length(model3$type, 3)
expect_identical(model3$cons$convergence$converged, TRUE)
expect_identical(model3$cons$convergence$criterion, 0)
expect_identical(model3$cons$convergence$difference, 0)
expect_identical(model3$cons$iterations$setting, 5)
expect_identical(model3$cons$iterations$actual, 2)
})
test_that("outlier_detection() works with iis_init(), tests turned on", {
skip_on_cran()
# base setup
set.seed(50)
p <- generate_param(dx1 = 1, dx2 = 1, dz2 = 2, intercept = TRUE,
beta = c(2, 4), sigma = 1, mean_z = matrix(c(0,0), 2, 1),
cov_z = matrix(c(1,0,0,1), 2, 2),
Sigma2_half = matrix(1), Omega2 = matrix(3/4),
Pi = t(matrix(c(1, 0, 0, 1, 0, 1), nrow = 2)))
d <- generate_data(parameters = p, n = 50)$data
formula <- y ~ -1+x1+x2 | -1+x1+z2+z3
gamma <- 0.05
# checked by stepping through function call that weak instrument test is actually applied
# iis arguments, no testing
arglist1 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = NULL)
model1 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist1)
# outliers: m0 = 5, m1 = 4, m2 = 2, m3 = 2, m4 = 2, m5 = 2
# add weak instrument test, should not pass
arglist2 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = 0.00000000001)
expect_warning(expect_error(outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist2),
"IIS final model is NULL. See warning."))
# add overid test, should not pass
arglist3 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = 0.99, weak = NULL)
expect_warning(expect_error(outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist3),
"IIS final model is NULL. See warning."))
# add both tests
arglist4 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = NULL,
turbo = FALSE, overid = 0.99, weak = 0.01)
expect_warning(expect_error(outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist4),
"IIS final model is NULL. See warning."))
# try the built-in tests from gets
# normality test
arglist5 <- list(t.pval = 0.01, do.pet = FALSE, normality.JarqueB = 0.05,
turbo = FALSE, overid = NULL, weak = NULL)
model5 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist5)
# differs from model1 b/c of testing
# one indicator retained in iter 0, and then no further outlier detected in iterations
# check this manually
manual5 <- ivgets::ivisat(formula = formula, data = d, iis = TRUE,
t.pval = gamma, normality.JarqueB = 0.05,
print.searchinfo = FALSE)
# only second block is selected over, first one probably fails (is issue in gets)
# one retained in second block (39)
# check that diagnostics are present, normality test has been used
expect_identical(rownames(manual5$selection$diagnostics),
c("Ljung-Box AR(1)", "Ljung-Box ARCH(1)", "Jarque-Bera"))
sel5 <- rep(TRUE, times = 50)
sel5[39] <- FALSE
names(sel5) <- as.character(1:50)
expect_identical(model5$sel$m0, sel5)
# PET test
arglist6 <- list(t.pval = 0.0000001, do.pet = TRUE, normality.JarqueB = NULL,
turbo = FALSE, overid = NULL, weak = NULL)
model6 <- outlier_detection(data = d, formula = formula, ref_dist = "normal",
sign_level = gamma, initial_est = "iis",
user_model = NULL, iterations = 5,
convergence_criterion = NULL, max_iter = NULL,
shuffle = FALSE, shuffle_seed = NULL, split = 0.5,
verbose = FALSE, iis_args = arglist6)
expect_snapshot_output(model1)
expect_snapshot_output(model5)
expect_snapshot_output(model6)
})
Try the robust2sls package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.