kerdec: Kernel deconvolution methods

# 
# library(kerdec)
# 
# ## Seed
# set.seed(666)
# 
# ## Settings and samples
# n <- 150                                # Sample size
# l <- 5                                  # Number of columns
# m <- n + 10                             # Error sample size
# shape <- 5                              # X distr. shape par.
# rate <- 5                               # X distr. rate par.
# sd_error <- .2                          # std. error of error distr.
# X <- rgamma(n, shape, rate)             # Uncontaminated sample
# eps_panel <- matrix(rlaplace(n*l, sd = sd_error),
#                     n, l)               # Panel of errors
# eps <- rlaplace(m, sd = sd_error)       # Pure errors
# Y <- X + eps_panel[, 1]                 # Contaminated sample
# Y_panel <- sweep(x = eps_panel, MARGIN = 1,
#                  STATS = X, FUN = "+")  # Contaminated in panel
# 
# 
# ## Real density
# plot(function(x) dgamma(x, shape, rate), -2, 6)
# #hist(Y)
# 
# ## Parameters to perform estimation
# lower <- -2
# upper <- 7
# 
# ## Case 1: normal error with known variance (misspecified
# ## distribution)
# case1 <-
#     kerdec_dens(Y, method = "NR", kernel = "triw", h0 = c(.06, 0.10),
#                 lower = lower, upper = upper, h = 0.08,
#                 error_dist = "normal",
#                 error_scale_par = sd_error)
# 
# with(case1, plot(x, f_vals, type = "l"))
# 
# ## Case 2: Laplace error with known variance
# case2 <-
#     kerdec_dens(Y, method = "NR", kernel = "triw", h0 = c(.05, 0.1),
#                 lower = lower, upper = upper, h = 0.068,
#                 error_dist = "Laplace",
#                 error_scale_par = sd_error)
# with(case2, plot(x, f_vals, type = "l"))
# 
# ## Case 3: normal error with unknown variance and sample of errors
# ## given (misspecified distribution)
# case3 <-
#     kerdec_dens(Y, method = "CV", kernel = "flat",
#                 lower = lower, upper = upper, h = 0.2,
#                 error_dist = "Normal", error_smp = eps)
# with(case3, plot(x, f_vals, type = "l"))
# 
# ## Case 4: Laplace error with known variance and sample of errors
# ## given
# case4 <-
#     kerdec_dens(Y, method = "CV", kernel = "flat",
#                 lower = lower, upper = upper, h = 0.2,
#                 error_dist = "Laplace", error_smp = eps)
# with(case4, plot(x, f_vals, type = "l"))
# 
# ## Case 4: ecf to approximate errors based on sample of errors
# ## given
# case5 <-
#     kerdec_dens(Y, method = "CV", kernel = "flat",
#                 lower = lower, upper = upper, h = 0.2,
#                 error_smp = eps)
# with(case5, plot(x, f_vals, type = "l"))
# 
# ## Case 6: Panel data with normal errors (unknown variances)
# case6 <- 
#   kerdec_dens(Y_panel, method = "CV", kernel = "flat",
#               lower = lower, upper = upper, h = 0.15, 
#               error_dist = "normal")
# with(case6, plot(x, f_vals, type = "l"))
# 
# ## Case 7: Panel data with Laplace errors (unknown variances)
# case7 <- 
#   kerdec_dens(smp = Y_panel, method = "CV", kernel = "flat",
#               lower = lower, upper = upper, h = 0.15, 
#               error_dist = "laplace")
# with(case7, plot(x, f_vals, type = "l"))
# 
# ## Case 8: Panel data with Laplace errors (unknown variances)
# case8 <- 
#   kerdec_dens(smp = Y_panel, method = "CV", kernel = "flat",
#               lower = lower, upper = upper, h = 0.15, 
#               error_dist = "none")
# with(case8, plot(x, f_vals, type = "l"))
# 
# ## Case 9: Panel data with normal errors (unknown variances)
# case9 <- 
#   kerdec_dens(Y_panel, method = "CV", kernel = "flat",
#               lower = lower, upper = upper, h = 0.15, 
#               error_dist = "normal", panel_proc = "take_aver")
# with(case9, plot(x, f_vals, type = "l"))
# 
# ## Case 10: Panel data with Laplace errors (unknown variances)
# case10 <- 
#   kerdec_dens(smp = Y_panel, method = "CV", kernel = "flat",
#               lower = lower, upper = upper, h = 0.15, 
#               error_dist = "laplace")
# with(case10, plot(x, f_vals, type = "l"), panel_proc = "take_aver")
# 
# ## Case 11: Panel data with Laplace errors (unknown variances)
# case11 <- 
#   kerdec_dens(smp = Y_panel, method = "CV", kernel = "flat",
#               lower = lower, upper = upper, h = 0.15, 
#               error_dist = "none")
# with(case11, plot(x, f_vals, type = "l"), panel_proc = "take_aver")
# 
# 
# 
# case0 <-
#   kerdec_dens(Y, method = "NR", kernel = "triw",
#               lower = lower, upper = upper, h = 0.07,
#               h0 = c(0.03, 0.2),
#               ## h0 = c(.04, 0.15),
#               error_dist = "laplace",
#               error_scale_par = sd_error)
# 
# with(case0, plot(x, f_vals, type = "l"))
# 
# kerdec_dens(Y, method = "NR", kernel = "triw",
#             lower = lower, upper = upper, h = 0.07,
#             h0 = c(0.05, 0.2),
#             ## h0 = c(.04, 0.15),
#             error_dist = "normal",
#             error_scale_par = sd_error)