Portable Interactive Control Charts • controlcharts

Introduction

A key limitation for interactive plots/charts in R is the need for a Shiny server in order to handle the data processing and statistical calculations. The controlcharts package overcomes this by performing all calculations in JavaScript, allowing the charts to be fully portable and interactive without the need for a server. By constructing charts using crosstalk data objects, multiple charts can also be linked together to allow for interactive filtering and highlighting in a fully portable format.

The controlcharts package also performs all necessary data aggregations and pre-processing in JavaScript. This means that users do not need to pre-aggregate or manipulate their data before creating charts, simplifying the workflow and making it easier to create interactive visualisations directly from raw data. This also allows for dynamic data exploration, as users can filter and interact with the charts without needing to re-run R code.

Generate Data

Consider a dataset containing monthly numerators and denominators for 10 different organisations over a 2-year period:

library(controlcharts)

set.seed(12324)

dat <- do.call(
  rbind.data.frame,
  lapply(toupper(letters[1:10]), function(grp) {
    denom <- sample(0:100, 24)
    data.frame(
      country = sample(c("C1", "C2", "C3"), 1, replace = TRUE),
      organisation = grp,
      month_start = seq(as.Date("2024-01-01"), length.out = 24, by = "month"),
      numerators = rbinom(n = 24, size = denom, prob = 0.3),
      denominators = denom
    )
  })
)
DT::datatable(dat)

Basic Interactivity

SPC charts and funnel plots can be constructed from this data directly, with the resulting HTML plots supporting mouseover tooltips:

spc_plt <- controlcharts::spc(data = dat,
                              keys = month_start,
                              numerators = numerators,
                              denominators = denominators,
                              spc_setting = list(chart_type = "p"),
                              outlier_settings = list(astronomical = TRUE,
                                                      shift = TRUE,
                                                      two_in_three = TRUE),
                              nhs_icon_settings =
                                list(show_variation_icons = TRUE))

fun_plt <- controlcharts::funnel(data = dat,
                                 keys = organisation,
                                 numerators = numerators,
                                 denominators = denominators)


spc_plt$html_plot

fun_plt$html_plot

Cross-plot Reactivity and Dynamic Calculations: Crosstalk

By converting the data to a crosstalk::SharedData object, multiple charts can be linked together to allow for interactive filtering and highlighting. As the aggregations, filtering, and statistical calculations are performed dynamically in JavaScript, these charts remain fully portable and interactive without the need for a Shiny server.

# Convert data to crosstalk SharedData object
crosstalk_dat <- crosstalk::SharedData$new(dat)

# Create charts using crosstalk data
spc_plt <- controlcharts::spc(data = crosstalk_dat,
                              keys = month_start,
                              numerators = numerators,
                              denominators = denominators,
                              spc_setting = list(chart_type = "p"),
                              outlier_settings = list(astronomical = TRUE,
                                                      shift = TRUE,
                                                      two_in_three = TRUE),
                              nhs_icon_settings =
                                list(show_variation_icons = TRUE))

fun_plt <- controlcharts::funnel(data = crosstalk_dat,
                                 keys = organisation,
                                 numerators = numerators,
                                 denominators = denominators)

We can then use crosstalk filters to allow users to dynamically filter the data displayed in both charts. As the filters are adjusted, the data aggregations and statistical calculations are updated in real-time within the charts:

country_filter <- crosstalk::filter_select("countryFilter", "Country",
                                           crosstalk_dat, ~country)
org_filter <- crosstalk::filter_select("orgFilter", "Organisation",
                                       crosstalk_dat, ~organisation)
date_filter <- crosstalk::filter_slider("monthFilter", "Date",
                                        crosstalk_dat, ~month_start)
crosstalk::bscols(
  date_filter,
  org_filter,
  country_filter
)
crosstalk::bscols(
  fun_plt$html_plot,
  spc_plt$html_plot
)

Date

Country

Organisation