Merge pull request #90 from pythonhealthdatascience/dev

Dev

Author: amyheather

DESCRIPTION

@@ -4,8 +4,11 @@ Imports:
     diffobj,
     dplyr,
     fitdistrplus,
+    future,
+    future.apply,
     ggplot2,
     jsonlite,
+    kableExtra,
     lintr,
     lubridate,
     plotly,

_quarto.yml

@@ -50,8 +50,8 @@ website:
       - pages/output_analysis/warmup.qmd
       - pages/output_analysis/outputs.qmd
       - pages/output_analysis/replications.qmd
-      - pages/output_analysis/n_reps.qmd
       - pages/output_analysis/parallel.qmd
+      - pages/output_analysis/n_reps.qmd
     - section: "Experimentation"
       contents:
       - pages/experiments/scenarios.qmd

environment.yaml

@@ -4,6 +4,7 @@ channels:
 dependencies:
   - black=25.1.0
   - flake8=7.3.0
+  - itables=2.5.2
   - jupyter=1.1.1
   - pandas=2.3.1
   - plotly=6.3.0

pages/model/patients.qmd

@@ -250,7 +250,7 @@ There are two arguments required by the initialiser: `param` and `run_number`.
 
 The `param` argument accepts an instance of the `Parameters` object, which groups all necessary simulation values in one place. This minimises the number of arguments required by `Model`, as it only needs one input for parameters.
 
-The `run_number` will be used when generating the random seeds. This is important when performing multiple replications, as it ensures each run gets a unique seed (see the [Replications](../output_analysis/n_reps.qmd) page for more details).
+The `run_number` will be used when generating the random seeds. This is important when performing multiple replications, as it ensures each run gets a unique seed (see the [Replications](../output_analysis/replications.qmd) page for more details).
 
 <br>
 
@@ -455,7 +455,7 @@ There are two arguments required by the function: `param` and `run_number`.
 
 The `param` argument requires the named list of parameters from the `create_params()` function, which groups all necessary simulation values in one list. This minimise the number of arguments required by `model()`, as it only needs one input for parameters.
 
-The `run_number` will be used when generating the random seeds. This is important when performing multiple replications, as it ensures each run gets a unique seed (see the [Replications](../output_analysis/n_reps.qmd) page for more details).
+The `run_number` will be used when generating the random seeds. This is important when performing multiple replications, as it ensures each run gets a unique seed (see the [Replications](../output_analysis/replications.qmd) page for more details).
 
 <br>
 

pages/output_analysis/outputs.qmd

@@ -45,6 +45,7 @@ from sim_tools.distributions import Exponential
 ```{r}
 #| output: false
 library(dplyr)
+library(knitr)
 library(simmer)
 library(tidyr)
 ```
@@ -147,7 +148,7 @@ We need to update the `Model` class so its `run` method returns a list of patien
 
 ### Runner class
 
-We're also introducing a new `Runner` class, responsible for running simulations and performing results calculations. This keeps calculation logic separate from the core modelling code. Also, this class will be modified to perform multiple model runs on the [Replications](n_reps.qmd) page.
+We're also introducing a new `Runner` class, responsible for running simulations and performing results calculations. This keeps calculation logic separate from the core modelling code. Also, this class will be modified to perform multiple model runs on the [Replications](replications.qmd) page.
 
 The basic structure of `Runner` is as follows. It initialises with simulation parameters, runs the model for a single replication, and organises results into separate patient-level and run-level outputs.
 
@@ -175,7 +176,36 @@ No changes required.
 
 ```{r}
 #| echo: false
-{{< include outputs_resources/create_params.R >}}
+#' Generate parameter list.
+#'
+#' @param interarrival_time Numeric. Time between arrivals (minutes).
+#' @param consultation_time Numeric. Mean length of doctor's
+#'   consultation (minutes).
+#' @param number_of_doctors Numeric. Number of doctors.
+#' @param warm_up_period Numeric. Duration of the warm-up period (minutes).
+#' @param data_collection_period Numeric. Duration of the data collection
+#'   period (minutes).
+#' @param verbose Boolean. Whether to print messages as simulation runs.
+#'
+#' @return A named list of parameters.
+
+create_params <- function(
+  interarrival_time = 5L,
+  consultation_time = 10L,
+  number_of_doctors = 3L,
+  warm_up_period = 30L, 
+  data_collection_period = 40L, 
+  verbose = FALSE
+) {
+  list(
+    interarrival_time = interarrival_time,
+    consultation_time = consultation_time,
+    number_of_doctors = number_of_doctors,
+    warm_up_period = warm_up_period, 
+    data_collection_period = data_collection_period, 
+    verbose = verbose
+  )
+}
 ```
 
 ### Warm-up function
@@ -184,7 +214,55 @@ No changes required.
 
 ```{r}
 #| echo: false
-{{< include outputs_resources/filter_warmup.R >}}
+#' Filters arrivals and resources to remove warm-up results.
+#'
+#' @param result Named list with two tables: monitored arrivals & resources.
+#' @param warm_up_period Length of warm-up period.
+#'
+#' @importFrom dplyr ungroup arrange slice n filter
+#'
+#' @return The name list `result`, but with the tables (`arrivals` and
+#' `resources`) filtered to remove warm-up results.
+#' @export
+
+filter_warmup <- function(result, warm_up_period) {
+  # Skip filtering if the warm-up period is zero
+  if (warm_up_period == 0L) return(result)
+
+  # Arrivals: Keep only patients who came during the data collection period
+  result[["arrivals"]] <- result[["arrivals"]] |>
+    group_by(.data[["name"]]) |>
+    filter(all(.data[["start_time"]] >= warm_up_period)) |>
+    ungroup()
+
+  # Resources: Filter to resource events in the data collection period
+  dc_resources <- filter(result[["resources"]],
+                         .data[["time"]] >= warm_up_period)
+
+  if (nrow(dc_resources) > 0L) {
+
+    # For each resource, get the last even during warm-up, and replace time
+    # for that event with the start time of the data collection period
+    last_usage <- result[["resources"]] |>
+      filter(.data[["time"]] < warm_up_period) |>
+      arrange(.data[["time"]]) |>
+      group_by(.data[["resource"]]) |>
+      slice(n()) |>
+      mutate(time = warm_up_period) |>
+      ungroup()
+
+    # Set that last event of the resource/s as the first row of the dataframe
+    # else calculations would assume all resources were idle at the start of
+    # the data collection period
+    result[["resources"]] <- rbind(last_usage, dc_resources)
+
+  } else {
+    # No events after warm-up; use filtered resource table as is
+    result[["resources"]] <- dc_resources
+  }
+
+  result
+}
 ```
 
 ### Run results function
@@ -226,9 +304,9 @@ We also need to correct the "replication" column. This is generated by `get_mon_
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["arrivals"]] |> arrange(start_time))
-knitr::kable(result[["resources"]])
-knitr::kable(result[["run_results"]])
+kable(result[["arrivals"]] |> arrange(start_time))
+kable(result[["resources"]])
+kable(result[["run_results"]])
 ```
 
 At the moment, `result[["run_results"]]` is pretty empty (just has the replication number). Let's add some performance measures!
@@ -390,8 +468,8 @@ In `arrivals` we can see six patients listed, which aligns with our results tabl
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["arrivals"]] |> arrange(start_time))
-knitr::kable(result[["run_results"]])
+kable(result[["arrivals"]] |> arrange(start_time))
+kable(result[["run_results"]])
 ```
 
 :::
@@ -778,7 +856,7 @@ model <- function(param, run_number) {
 
   # Get the extra arrivals attributes#<<
   extra_attributes <- get_mon_attributes(env) |>#<<
-    select("name", "key", "value") |>#<<
+    dplyr::select("name", "key", "value") |>#<<
     # Add column with resource name, and remove that from key#<<
     mutate(resource = gsub("_.+", "", .data[["key"]]),#<<
            key = gsub("^[^_]+_", "", .data[["key"]])) |>#<<
@@ -814,8 +892,8 @@ model <- function(param, run_number) {
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["arrivals"]] |> arrange(start_time))
-knitr::kable(result[["run_results"]])
+kable(result[["arrivals"]] |> arrange(start_time))
+kable(result[["run_results"]])
 ```
 
 :::
@@ -1031,7 +1109,7 @@ model <- function(param, run_number) {
 
   # Get the extra arrivals attributes#<<
   extra_attributes <- get_mon_attributes(env) |>#<<
-    select("name", "key", "value") |>#<<
+    dplyr::select("name", "key", "value") |>#<<
     # Add column with resource name, and remove that from key#<<
     mutate(resource = gsub("_.+", "", .data[["key"]]),#<<
            key = gsub("^[^_]+_", "", .data[["key"]])) |>#<<
@@ -1065,8 +1143,8 @@ model <- function(param, run_number) {
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["arrivals"]] |> arrange(start_time))
-knitr::kable(result[["run_results"]])
+kable(result[["arrivals"]] |> arrange(start_time))
+kable(result[["run_results"]])
 ```
 
 :::
@@ -1930,7 +2008,7 @@ No changes required.
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["run_results"]])
+kable(result[["run_results"]])
 ```
 
 <br>
@@ -1940,7 +2018,7 @@ We can also view the result with `summarise = FALSE`:
 ```{r}
 util_df <- calc_utilisation(resources = result[["resources"]],
                             summarise = FALSE)
-knitr::kable(util_df)
+kable(util_df)
 ```
 
 :::
@@ -2320,8 +2398,8 @@ model <- function(param, run_number) {
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["arrivals"]] |> arrange(start_time))
-knitr::kable(result[["run_results"]])
+kable(result[["arrivals"]] |> arrange(start_time))
+kable(result[["run_results"]])
 ```
 
 :::
@@ -2717,8 +2795,8 @@ model <- function(param, run_number) {
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["arrivals"]] |> arrange(start_time))
-knitr::kable(result[["run_results"]])
+kable(result[["arrivals"]] |> arrange(start_time))
+kable(result[["run_results"]])
 ```
 
 :::
@@ -3175,8 +3253,8 @@ Here, we see the new `result[["patient_in_system"]]` dataframe, as well as the r
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["patients_in_system"]])
-knitr::kable(result[["run_results"]])
+kable(result[["patients_in_system"]])
+kable(result[["run_results"]])
 ```
 
 :::
@@ -3688,7 +3766,7 @@ model <- function(param, run_number) {
 
   # Get the extra arrivals attributes#<<
   extra_attributes <- get_mon_attributes(env) |>#<<
-    select("name", "key", "value") |>#<<
+    dplyr::select("name", "key", "value") |>#<<
     # Add column with resource name, and remove that from key#<<
     mutate(resource = gsub("_.+", "", .data[["key"]]),#<<
            key = gsub("^[^_]+_", "", .data[["key"]])) |>#<<
@@ -3729,8 +3807,8 @@ model <- function(param, run_number) {
 ```{r}
 param <- create_params()
 result <- model(param = param, run_number = 1L)
-knitr::kable(result[["arrivals"]])
-knitr::kable(result[["run_results"]])
+kable(result[["arrivals"]])
+kable(result[["run_results"]])
 ```
 
 :::