Pipeline: Pipeline

In dymium-org/dymiumCore: A Toolkit for Building a Dynamic Microsimulation Model for Integrated Urban Modelling

Description

Pipeline acts as a wraper for event functions. It can shuffle the order in which the wrapped event functions get executed and also control how many times they should be run within the pipeline. The later capability allows the modeller to mix events with different time resolutions in a single microsimulation pipeline. A more tangible example would be when you are modelling residential relocation of households, renters may relocate every 6 months but buyers don't move that often hence you can use the Pipeline to run a renter relocation event twice for every run of a buy relocation event.

pipeline <- Pipeline$new(. %>%
                          renter_relocation_event())

for (year in 1:5) {
  world %>%
    pipeline$run(n_loops = 2) %>%
    buyer_relocation_event()
}

Format

R6::R6Class object inheriting from Generic.

Construction

Pipeline$new(x)

• x
  Event functions in a %>% structure where the start of the pipe must be .. \ See the example section below.

Public Methods

• set(x)
  Sets the event functions inside the Pipeline object.

• get()
  Returns the event functions.

• run(x, shuffle = FALSE, n_loops = 1L)
  (Container, logical(1), integer(1)) -> Container
  Execute the event functions in the order that it was added. To make the order randomised set shuffle as TRUE. n_loops controls how many times should the event functions be executed before exiting the Pipeline. If it is greater than one then the event functions will be run multiple times it the same order or randomised orders each time, depending on the value in shuffle.

• print()
  

Examples

# create Population
Pop <- Population$new(toy_individuals, toy_households, pid_col = "pid", hid_col = "hid")

# create 3 dummy events
event1 <- function(object) {
  # do something
  return(object)
}

event2 <- function(object) {
  # do something
  return(object)
}

event3 <- function(object) {
  # do something
  return(object)
}

# construct the sequence of the dummy events
# pipeline <- Pipeline$new(. %>% event1 %>% event2 %>% event3)
# pipeline$get()

# run events in a random order
# pipeline$run(x = Pop, shuffle = TRUE)

dymium-org/dymiumCore documentation built on July 18, 2021, 5:10 p.m.