Add the code used for generating synthetic_data_for_UBA

Static documentation except articles rebuilt by pkgdown

1 files changed, 79 insertions, 3 deletions

diff --git a/man/synthetic_data_for_UBA.Rd b/man/synthetic_data_for_UBA.Rd
index a9739d55..3339798e 100644
--- a/man/synthetic_data_for_UBA.Rd
+++ b/man/synthetic_data_for_UBA.Rd
@@ -11,7 +11,7 @@
 

  Variance component 'a' is based on a normal distribution with standard deviation of 3,

  Variance component 'b' is also based on a normal distribution, but with a standard deviation of 7.

- Variance component 'c' is based on the error model from Rocke and Lorenzato (1995), with the 

+ Variance component 'c' is based on the error model from Rocke and Lorenzato (1995), with the

  minimum standard deviation (for small y values) of 0.5, and a proportionality constant of 0.07

  for the increase of the standard deviation with y.

 

@@ -36,12 +36,13 @@
 \source{

   Ranke (2014) Prüfung und Validierung von Modellierungssoftware als Alternative

   zu ModelMaker 4.0, Umweltbundesamt Projektnummer 27452

- 

+

   Rocke, David M. und Lorenzato, Stefan (1995) A two-component model for

   measurement error in analytical chemistry. Technometrics 37(2), 176-184.

 }

 \examples{

 \dontrun{

+# The data have been generated using the following kinetic models

 m_synth_SFO_lin <- mkinmod(parent = list(type = "SFO", to = "M1"),

                            M1 = list(type = "SFO", to = "M2"),

                            M2 = list(type = "SFO"), use_of_ff = "max")

@@ -61,7 +62,82 @@ m_synth_DFOP_par <- mkinmod(parent = list(type = "DFOP", to = c("M1", "M2"),
                             M1 = list(type = "SFO"),

                             M2 = list(type = "SFO"), use_of_ff = "max")

 

-fit <- mkinfit(m_synth_SFO_lin, synthetic_data_for_UBA_2014[[1]]$data, quiet = TRUE)

+# The model predictions without intentional error were generated as follows

+sampling_times = c(0, 1, 3, 7, 14, 28, 60, 90, 120)

+

+d_synth_SFO_lin <- mkinpredict(m_synth_SFO_lin,

+                               c(k_parent = 0.7, f_parent_to_M1 = 0.8,

+                                 k_M1 = 0.3, f_M1_to_M2 = 0.7,

+                                 k_M2 = 0.02),

+                               c(parent = 100, M1 = 0, M2 = 0),

+                               sampling_times)

+

+d_synth_DFOP_lin <- mkinpredict(m_synth_DFOP_lin,

+                                c(k1 = 0.2, k2 = 0.02, g = 0.5,

+                                  f_parent_to_M1 = 0.5, k_M1 = 0.3,

+                                  f_M1_to_M2 = 0.7, k_M2 = 0.02),

+                                 c(parent = 100, M1 = 0, M2 = 0),

+                                 sampling_times)

+

+d_synth_SFO_par <- mkinpredict(m_synth_SFO_par,

+                               c(k_parent = 0.2,

+                                 f_parent_to_M1 = 0.8, k_M1 = 0.01,

+                                 f_parent_to_M2 = 0.2, k_M2 = 0.02),

+                                 c(parent = 100, M1 = 0, M2 = 0),

+                                 sampling_times)

+

+d_synth_DFOP_par <- mkinpredict(m_synth_DFOP_par,

+                               c(k1 = 0.3, k2 = 0.02, g = 0.7,

+                                 f_parent_to_M1 = 0.6, k_M1 = 0.04,

+                                 f_parent_to_M2 = 0.4, k_M2 = 0.01),

+                                 c(parent = 100, M1 = 0, M2 = 0),

+                                 sampling_times)

+

+# Construct names for datasets with errors

+d_synth_names = paste0("d_synth_", c("SFO_lin", "SFO_par",

+                                     "DFOP_lin", "DFOP_par"))

+

+# Function for adding errors. The add_err function now published with this

+# package is a slightly generalised version where the names of secondary

+# compartments that should have an initial value of zero (M1 and M2 in this

+# case) are not hardcoded any more.

+add_err = function(d, sdfunc, LOD = 0.1, reps = 2, seed = 123456789)

+{

+  set.seed(seed)

+  d_long = mkin_wide_to_long(d, time = "time")

+  d_rep = data.frame(lapply(d_long, rep, each = 2))

+  d_rep$value = rnorm(length(d_rep$value), d_rep$value, sdfunc(d_rep$value))

+

+  d_rep[d_rep$time == 0 & d_rep$name %in% c("M1", "M2"), "value"] <- 0

+  d_NA <- transform(d_rep, value = ifelse(value < LOD, NA, value))

+  d_NA$value <- round(d_NA$value, 1)

+  return(d_NA)

+}

+

+# The following is the two-component model of Rocke and Lorenzato (1995)

+sdfunc_twocomp = function(value, sd_low, rsd_high) {

+  sqrt(sd_low^2 + value^2 * rsd_high^2)

+}

+

+# Add the errors.

+for (d_synth_name in d_synth_names)

+{

+  d_synth = get(d_synth_name)

+  assign(paste0(d_synth_name, "_a"), add_err(d_synth, function(value) 3))

+  assign(paste0(d_synth_name, "_b"), add_err(d_synth, function(value) 7))

+  assign(paste0(d_synth_name, "_c"), add_err(d_synth,

+                           function(value) sdfunc_twocomp(value, 0.5, 0.07)))

+

+}

+

+d_synth_err_names = c(

+  paste(rep(d_synth_names, each = 3), letters[1:3], sep = "_")

+)

+

+# This is just one example of an evaluation using the kinetic model used for

+# the generation of the data

+fit <- mkinfit(m_synth_SFO_lin, synthetic_data_for_UBA_2014[[1]]$data,

+               quiet = TRUE)

 plot_sep(fit)

 summary(fit)

 }