8-bit-sheep / googleAnalyticsR

Use the Google Analytics API from R
https://8-bit-sheep.com/googleAnalyticsR/
Other
257 stars 76 forks source link

Create Shiny apps via template from ga_model objects #358

Closed MarkEdmondson1234 closed 2 weeks ago

MarkEdmondson1234 commented 3 years ago

Include and allow shiny templates that work with pre-defined ga_model plotting.

MarkEdmondson1234 commented 3 years ago

Weird shiny module behaviour - looks like you need to refresh the model each time? Loading from file doesn't seem to work, breaks references or something.

library(googleAnalyticsR)
ga_model_shiny(ga_model_example("ga4-trend.gamr"), template = ga_model_shiny_template("shinydashboard_ga4"), skin="blue")
# Error in shiny::NS(id) : argument "id" is missing, with no default

mo <- ga_model_example("ga4-trend.gamr")
ga_model_shiny(mo, template = ga_model_shiny_template("shinydashboard_ga4"), skin = "blue")
# Error in shiny::NS(id) : argument "id" is missing, with no default

Recreate module from https://github.com/MarkEdmondson1234/googleAnalyticsR/issues/354#issuecomment-753400851

# fetch data
data_f <- function(view_id,date_range = c("400daysAgo","yesterday"),metrics = c("sessions"),
...)
{

ga_data(view_id, metrics = metrics, date_range = date_range, dimensions = "date", limit = -1, orderBys = ga_data_order(+date))

}

# model data
model_f <- function(df,
...)
{

xts::xts(df[, -1], order.by = df$date)

}

# output data
output_f<- function(df,...)
{
require(dygraphs)
dygraph(df, main = "GA4 trend- googleAnalyticsR") %>% 
    dyAxis("x", label = "Date") %>%
    dyOptions(axisLineWidth = 1.5, drawGrid = FALSE)

}

model <- ga_model_make(data_f = data_f, required_columns = c("date"), model_f = model_f, output_f = output_f, required_packages = c("xts","dygraphs"), description = "GA4 Metric Trend", outputShiny = dygraphs::dygraphOutput, renderShiny = dygraphs::renderDygraph)

Now it works

ga_model_shiny(model, template = ga_model_shiny_template("shinydashboard_ga4"), skin="blue")

This works

ga_model_save(model, "test.gamr")
ga_model_shiny(ga_model_load("test.gamr"), template = ga_model_shiny_template("shinydashboard_ga4"), skin="blue")

So try saving it to the package..

ga_model_save(model, "inst/models/ga4-trend.gamr")
# works
ga_model_shiny(ga_model_load("inst/models/ga4-trend.gamr"), template = ga_model_shiny_template("shinydashboard_ga4"), skin="blue")

Build package, restart R - still all work.

ga_model_shiny(ga_model_load("inst/models/ga4-trend.gamr"), template = ga_model_shiny_template("shinydashboard_ga4"), skin="blue")

ga_model_shiny(ga_model_example("ga4-trend.gamr"), template = ga_model_shiny_template("shinydashboard_ga4"), skin="blue")

ga_model_shiny(ga_model_load("test.gamr"), template = ga_model_shiny_template("shinydashboard_ga4"), skin="blue")

... will comment again when it happens.....

MarkEdmondson1234 commented 3 years ago

A weird warning appears when loading shiny models: Warning in serialize(object, connection = NULL, ascii = ascii, version = serializeVersion) : 'package:stats' may not be available when loading

MarkEdmondson1234 commented 3 years ago

Added a way to also have module inputs in the rendered Shiny models, as well as multiple models in one template.


library(CausalImpact)
library(xts)
library(tidyr)
library(googleAnalyticsR)
library(assertthat)
library(dygraphs)

# fetch data
data_f <- function(view_id, date_range = c(Sys.Date() - 600, Sys.Date()), ...) {
    google_analytics(view_id, date_range = date_range, metrics = "sessions", dimensions = c("date", 
        "channelGrouping"), max = -1)
}

# model data
model_f <- function(df, event_date, response = "Organic Search", predictors = c("Video", 
    "Social", "Direct"), ...) {
    message("CausalImpact input data columns: ", paste(names(df), collapse = " "))
    stopifnot(is.character(response), length(response) == 1, assertthat::is.date(event_date), 
        is.character(predictors))
    pivoted <- df %>% tidyr::spread(channelGrouping, sessions)
    stopifnot(response %in% names(pivoted))
    web_data_xts <- xts::xts(pivoted[-1], order.by = as.Date(pivoted$date), frequency = 7)
    pre.period <- as.Date(c(min(df$date), event_date))
    post.period <- as.Date(c(event_date + 1, max(df$date)))
    predictors <- intersect(predictors, names(web_data_xts))
    model_data <- web_data_xts[, c(response, predictors)]
    names(model_data) <- make.names(names(model_data))
    model_data[is.na(model_data)] <- 0
    CausalImpact::CausalImpact(model_data, pre.period, post.period)
}

# output data
output_f <- function(impact, event_date, ...) {
    ci <- impact$series
    ci <- xts::xts(ci)
    dygraph(data = ci[, c("response", "point.pred", "point.pred.lower", "point.pred.upper")], 
        main = "Expected (95% confidence level) vs Observed", group = "ci") %>% dyEvent(x = event_date, 
        "Event") %>% dySeries(c("point.pred.lower", "point.pred", "point.pred.upper"), 
        label = "Expected") %>% dySeries("response", label = "Observed")
}

# shiny input function
uiInput <- shiny::dateInput("event_date", "Event Date", Sys.Date() - 30)

# use via ga_model_make()

ga_model_edit("inst/models/ga-effect.gamr", inputShiny = uiInput, data_f = data_f, model_f = model_f, output_f = output_f, outputShiny = dygraphs::dygraphOutput, renderShiny = dygraphs::renderDygraph)

ga_model_shiny("inst/models/ga-effect.gamr", template = "inst/models/shiny/template_ua.R")
MarkEdmondson1234 commented 3 years ago

Handle multiple inputShiny within one module e.g. look for all the x$attribs$ids and apply the shiny::NS() function to it, update the dots arguments. Need to extract id as its not consistent see https://github.com/rstudio/shiny/issues/3248

MarkEdmondson1234 commented 3 years ago

Working with multiple IDs now, which means templates can be more generic.

ga_model_shiny("inst/models/time-normalised.gamr", template = ga_model_shiny_template("template_ua.R"))
MarkEdmondson1234 commented 3 years ago
  m1 <- ga_model_example("decomp_ga.gamr")
  m2 <- ga_model_example("decomp_ga_advanced.gamr")

  # launch single shiny app
  ga_model_shiny(m1, template = ga_model_shiny_template("template_ua.R"))
  ga_model_shiny(m2, template = ga_model_shiny_template("template_ua.R"))

  # launch two models in one shiny app
  ga_model_shiny(list(m1,m2), 
                 template = ga_model_shiny_template("multiple_ua.R"))

  m3 <- ga_model_example("time-normalised.gamr")
  m4 <- ga_model_example("ga-effect.gamr")
  # launch in gentelella template
  ga_model_shiny(list(m4,m3), 
                 template = ga_model_shiny_template("gentelella.R"))
MarkEdmondson1234 commented 3 years ago
MarkEdmondson1234 commented 3 years ago
         reactive_dots <- shiny::reactive({
          copy_input_ids(input_ids, input, dots)
        })

        # 
        data_inputs <- shiny::reactive({
          data_args <- formals(f)
          data_args$view_id <- NULL

          dot_names <- reactive_dots()

          o <- lapply(names(data_args), function(x){
            if(x %in% names(dot_names)){
              return(data_args[[x]])
            } else {
              myMessage("isolating ", x)
              shiny::isolate(dot_names[[x]])
            }
          })
          setNames(o, names(data_args))
        })
MarkEdmondson1234 commented 3 years ago

Templates can now carry the model libraries, authentication dropdowns and load multiple models, which makes templates be able to be more generic and work for GA4 and Universal

ga_model_shiny(list(ga_model_example("decomp_ga.gamr"), ga_model_example("time-normalised.gamr")), auth_dropdown = "uni", template = ga_model_shiny_template("template1.R"))

Example template used above:

library(shiny)
library(googleAuthR)
library(googleAnalyticsR)
{{{ model_libraries }}}

gar_set_client(web_json = "{{ web_json }}",
               scopes = "{{ scopes }}")
options(googleAuthR.redirect = "{{ deployed_url }}")

# loads pre-existing models
{{{ model_load }}}

## ui.R
ui <- fluidPage(title = "{{ shiny_title }}",
                {{ auth_ui }},
                h2("Model Output"),
{{{ model_ui }}}

)

## server.R
server <- function(input, output, session){

  token <- gar_shiny_auth(session)

  {{{ auth_accounts }}}

  # module for authentication
  view_id <- {{ auth_server }}

  # module to display model results
{{{ model_server }}}

}

shinyApp(gar_shiny_ui(ui, login_ui = silent_auth), server)

Gentella theme

m3 <- ga_model_example("time-normalised.gamr")
m4 <- ga_model_example("ga-effect.gamr")
# launch in gentelella template
ga_model_shiny(list(m4,m3), auth_dropdown = "universal",
               template = ga_model_shiny_template("gentelella.R"))
library(shiny)             # R webapps
library(gentelellaShiny)   # ui theme
library(googleAuthR)       # auth login
library(googleAnalyticsR) # get google analytics
{{{ model_libraries }}}

# takes JSON client secrets from GAR_CLIENT_WEB_JSON
# set before calls to googleAnalyticsR to make sure it doesn't use default project.
gar_set_client(web_json = "{{ web_json }}",
               scopes = "{{ scopes }}")

options(googleAuthR.redirect = "{{ deployed_url }}")

# loads a pre-existing models, or NULL if they aren't present
{{{ model_load }}}

ui <- gentelellaPage(
  menuItems = sideBarElement(a("Start Again", href="/")),
  title_tag = "GA time normalised pages",
  site_title = a(class="site_title", icon("clock"), span("Time Normalised")),
  footer = "Made with googleAnalyticsR::ga_model_shiny()",

  # shiny UI elements
  h3("Choose GA account"),
  {{ auth_ui }},
  h3("Time Normalised pages"),
  {{{ model_ui }}},
  br()

)

server <- function(input, output, session) {

  token <- gar_shiny_auth(session)

  {{{ auth_accounts }}}

  # module for authentication
  view_id <- {{ auth_server }}

  # module to display model results
  {{{ model_server }}}

}
# Run the application
shinyApp(gar_shiny_ui(ui, login_ui = silent_auth), server)
MarkEdmondson1234 commented 3 years ago

It gets a bit ridic but you could just template the whole server.R and leave people to customise the ui.R bit only.

MarkEdmondson1234 commented 3 years ago

Support for custom wrapping of model UI output by supplying a function

# make a function to output the custom shinydashboard tabs
shinydashboard_ui_menu <- function(models){

  model_n <- paste0("model", seq_along(models)) 
  labels <- lapply(models, function(x) substr(x$description, 0,14))

  f <- function(model_n, label){
    paste(
      sprintf(
        "menuItem('%s', tabName = '%s')", 
        label, model_n
      ),
      collapse = ",\n"
    )}

  mapply(f, model_n, labels, SIMPLIFY = FALSE, USE.NAMES = FALSE)
}

# supply custom function for wrapping the model_ui output with tabItem()
shinydashboard_ui <- function(model_n){

  paste(
    sprintf(
      "tabItem(tabName = '%s',
         %s$ui('%s'))", 
      model_n, model_n, model_n
    ),
    collapse = ",\n"
  )
}

m3 <- ga_model_example("time-normalised.gamr")
m4 <- ga_model_example("ga-effect.gamr")
models <- list(m3, m4)

# launch shiny app with the models in each tab
# model_tabs is via ... and a custom macro in the shinydashboard template
ga_model_shiny(models, auth_dropdown = "universal", 
                            template = ga_model_shiny_template("shinydashboard.R"), 
                            ui_f = shinydashboard_ui, 
                            model_tabs = shinydashboard_ui_menu(models))
image
MarkEdmondson1234 commented 3 years ago

Can add boilerplate to the templates so end user templates only need to make the UI which is very cool

e.g.

library(gentelellaShiny)   # ui theme

ui <- gentelellaPage(
  menuItems = sideBarElement(a("Start Again", href="/")),
  title_tag = "GA time normalised pages",
  site_title = a(class="site_title", icon("clock"), span("Time Normalised")),
  footer = "Made with googleAnalyticsR::ga_model_shiny()",

  # shiny UI elements
  h3("Choose GA account"),
  {{ auth_ui }},
  {{{ date_range }}},
  h3("Time Normalised pages"),
  {{{ model_ui }}},
  br()

)
MarkEdmondson1234 commented 3 years ago

May as well make it a ui.R and server.R file

MarkEdmondson1234 commented 3 years ago

Refactored to allow www themes, ui.R and folders all supported

# see Shiny templates included with the package
ga_model_shiny_template("list")

## Not run: 

# a universal analytics model using default template "basic"
ga_model_shiny(
  ga_model_example("decomp_ga.gamr"), 
  auth_dropdown = "universal")

# a template from a directory holding an app.R file
ga_model_shiny(
  ga_model_example("decomp_ga.gamr"), 
  auth_dropdown = "universal",
  template = ga_model_shiny_template("basic_app"))

# a template directly from an app.R file that has its own server object
ga_model_shiny(
  ga_model_example("decomp_ga.gamr"), 
  auth_dropdown = "universal",
  template = ga_model_shiny_template("basic_app/app.R"))

# a template from only an ui.R file that will import boilerplate server.R
ga_model_shiny(
  ga_model_example("decomp_ga.gamr"), 
  auth_dropdown = "universal",
  template = ga_model_shiny_template("basic/ui.R"))
MarkEdmondson1234 commented 3 years ago

Get a sweet theme running via https://shiny.rstudio.com/articles/templates.html