Examples

An MNL design

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
#' Specifying a utility function with 3 attributes and a constant for the
#' SQ alternative. The design has 20 rows.
utility <- list(
  alt1 = "b_x1[0.1]  * x1[1:5] + b_x2[0.4] * x2[c(0, 1)] + b_x3[-0.2] * x3[seq(0, 1, 0.25)]",
  alt2 = "b_x1       * x1      + b_x2      * x2          + b_x3       * x3",
  alt3 = "b_sq[0.15] * sq[1]"
)

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "rsc", draws = "scrambled-sobol",
                          control = list(
                            max_iter = 21000,
                            max_no_improve = 5000
                          ))

# Add a blocking variable to the design with 4 blocks.
design <- block(design, 4)


summary(design)

An MNL design with interactions

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
utility <- list(
  alt1 = "b_x1[0.1] * x1[1:5] + b_x2[0.4] * x2[c(0, 1)] + b_x3[-0.2] * x3[seq(0, 1, 0.25)] + b_x1x2[-0.1] * I(x1 * x2)",
  alt2 = "b_x1      * x1      + b_x2      * x2          + b_x3       * x3"
)

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "federov", draws = "scrambled-sobol",
                          dudx = "b_x3")

summary(design)

An MNL design with multiway interactions

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
utility <- list(
  alt1 = "b_x1[0.1] * x1[1:5] + b_x2[0.4] * x2[c(0, 1)] + b_x3[-0.2] * x3[seq(0, 1, 0.25)] + b_x1x2[-0.1] * I(x1 * x2)",
  alt2 = "b_x1      * x1      + b_x2      * x2          + b_x3       * x3 + b_x1x2x3[0.1] * I(x1 * x2 * x3)"
)

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "rsc", draws = "scrambled-sobol",
                          dudx = "b_x3")

summary(design)

An MNL design with dummy-coded attributes

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
#' A design where the first attribute is dummy-coded.
utility <- list(
  alt1 = "b_x1_dummy[c(0.1, 0.2)] * x1[c(1, 2, 3)] + b_x2[0.4] * x2[c(0, 1)] + b_x3[-0.2] * x3[seq(0, 1, 0.25)]",
  alt2 = "b_x1_dummy              * x1             + b_x2      * x2          + b_x3       * x3"
)

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "rsc", draws = "scrambled-sobol")

# Add a blocking variable to the design with 2 blocks.
design <- block(design, 2)


summary(design)

An MNL design with specified level occurrences

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
utility <- list(
  alt1 = "b_x1_dummy[c(0.1, 0.2)] * x1[c(1, 2, 3)](4:14) + b_x2[0.4] * x2[c(0, 1)](9:11) + b_x3[-0.2] * x3[seq(0, 1, 0.25)]",
  alt2 = "b_x1_dummy              * x1                   + b_x2      * x2                + b_x3       * x3"
)


# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "federov", draws = "scrambled-sobol")

design <- block(design, 4)

summary(design)

An MNL design with alternative specific attributes

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
utility <- list(
  car   = "                          b_travel_time[-0.1] * travel_time_car[c(10, 15, 20, 25)]   + b_travel_cost[-0.2] * travel_cost_car[c(25, 50, 75, 100)]  + b_comfort_dummy[c(0.1, 0.2)] * comfort[c(1, 2, 3)]",
  bus   = "b_bus[-0.1]  * bus[1]   + b_travel_time       * travel_time_bus[c(20, 25, 30, 35)]   + b_travel_cost       * travel_cost_bus[c(10, 15, 20, 25)]   + b_comfort_dummy              * comfort",
  train = "b_train[0.1] * train[1] + b_travel_time       * travel_time_train[c(15, 20, 25, 30)] + b_travel_cost       * travel_cost_train[c(20, 30, 40, 50)] + b_comfort_dummy              * comfort"
)

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "rsc", draws = "scrambled-sobol",
                          control = list(
                            max_iter = 21000,
                            efficiency_threshold = 0.01
                          ))

# Add a blocking variable to the design with 4 blocks.
design <- block(design, 4)


summary(design)

An MNL design with Bayesian priors

#
# Example file for creating a simple MNL design with Bayesian priors
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
#' Generating a design with Bayesian priors.
utility <- list(
  alt1 = "b_x1[0.1] * x1[2:5] + b_x2[uniform_p(-1, 1)] * x2[c(0, 1)] + b_x3[normal_p(0, 1)] * x3[seq(0, 1, 0.25)]",
  alt2 = "b_x1      * x1      + b_x2                   * x2          + b_x3                 * x3"
)

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "rsc", draws = "scrambled-sobol",
                          control = list(
                            max_iter = 10000
                          ))

summary(design)

An MNL design with dummy-coded attributes, Bayesian priors, and restrictions on level occurrence

#
# Example file for creating a simple MNL design with Bayesian priors
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
#' Generating a design with Bayesian priors using dummy coding and level occurrences
#' NOTE: This design may take a long time. It has to first find a candidate that
#' meets the restrictions and then evaluate whether it is better than the
#' current best. Little information is provided about the process along the way.
utility <- list(
  alt1 = "b_x1_dummy[c(uniform_p(-1, 1), uniform_p(-1, 1))] * x1[c(1, 2, 3)](6:10, 4:14, 6:10) + b_x2[0.4] * x2[c(0, 1)](9:11) + b_x3[-0.2] * x3[seq(0, 1, 0.25)]",
  alt2 = "b_x1_dummy                                        * x1                               + b_x2      * x2                + b_x3       * x3"
)


# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "random", draws = "scrambled-sobol",
                          control = list(
                            max_iter = 10000
                          ))

summary(design)

An MNL design with a supplied candidate set

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
utility <- list(
  alt1 = "b_x1[0.1] * x1[2:5] + b_x2[0.4] * x2[c(0, 1)] + b_x3[-0.2] * x3[seq(0, 1, 0.25)]",
  alt2 = "b_x1      * x1      + b_x2      * x2          + b_x3       * x3",
  alt3 = "b_sq[0]   * sq[1]"
)

# Use the full factorial as the candidate set
candidate_set <- full_factorial(
  list(
    alt1_x1 = 2:5,
    alt1_x2 = c(0, 1),
    alt1_x3 = seq(0, 1, 0.25),
    alt2_x1 = 2:5,
    alt2_x2 = c(0, 1),
    alt2_x3 = seq(0, 1, 0.25),
    alt3_sq = 1
  )
)

candidate_set <- candidate_set[!(candidate_set$alt1_x1 == 2 & candidate_set$alt1_x2 == 0 & candidate_set$alt1_x3 == 0), ]
candidate_set <- candidate_set[!(candidate_set$alt2_x2 == 1 & candidate_set$alt2_x3 == 1), ]

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "federov", draws = "scrambled-sobol",
                          candidate_set = candidate_set)


summary(design)

An MNL design with specified explusions applied to the candidate set

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
utility <- list(
  alt1 = "b_x1[0.1] * x1[2:5] + b_x2[0.4] * x2[c(0, 1)] + b_x3[-0.2] * x3[seq(0, 1, 0.25)]",
  alt2 = "b_x1      * x1      + b_x2      * x2          + b_x3       * x3"
)

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          model = "mnl", efficiency_criteria = "d-error",
                          algorithm = "federov", draws = "scrambled-sobol",
                          exclusions = list(
                            "alt1_x1 == 2 & alt1_x2 == 0 & alt1_x3 == 0",
                            "alt2_x2 == 1 & alt2_x3 == 1"
                          ))

An MNL design optimized for c-efficiency

#
# Example file for creating a simple MNL design
#
rm(list = ls(all = TRUE))
# library(spdesign)

# Define the list of utility functions ----
#' Specifying a utility function with 3 attributes and a constant for the
#' SQ alternative. The design has 20 rows.
utility <- list(
  alt1 = "b_x1[0.1]  * x1[1:5] + b_x2[0.4] * x2[c(0, 1)] + b_x3[-0.2] * x3[seq(0, 1, 0.25)]",
  alt2 = "b_x1       * x1      + b_x2      * x2          + b_x3       * x3",
  alt3 = "b_sq[0.15] * sq[1]"
)

# Generate designs ----
design <- generate_design(utility, rows = 20,
                          dudx = "b_x3",
                          model = "mnl",
                          efficiency_criteria = "c-error",
                          algorithm = "rsc",
                          draws = "scrambled-sobol",
                          control = list(
                            max_iter = 21000,
                            max_no_improve = 5000
                          ))

# Add a blocking variable to the design with 4 blocks.
design <- block(design, 4)


summary(design)