Building a Google Analytics Dashboard With r Shiny From Scratch – Part 1
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
I participated in the R Shiny 2021 contest and published an application similar to the Google Analytics dashboard app. For that, I used the Google Analytics API and the Google Search Console API to pull my own data from my blog directly into the application.
The application uses the shinyauthr
library because the dashboard can be adjusted for each user who has their own username and password. On the first page, there are some visualizations for page views, devices used, etc. On the second page, there is a time-series model that tries to predict my page views two months in advance.
I have been learning R Shiny seriously for a year now and tried to put in as much knowledge as possible from my learning over the past year. Some features of the app include:
- JavaScript and jQuery code + some CSS and Bootstrap classes for a better user experience
- The Shiny router package from Appsilon to make the app appear like a multi page application.
- A MongoDB database to save user data and individualize the dashboard.
- Modularization of code to create a maintainable application that is easy to extend.
- Custom functions for better unit testing.
- R Shiny testing.
- R Shiny application as a package for easier testing and better documentation.
- A time series model with
modeltime
This blog post series consists of multiple parts and we are coding the application together from scratch. Let’s get started.
Basic UI
The first part will be the hardest one. That is not to say that is hard, but it will be the hardest one for me because we are going to use some JavaScript and jQuery which I am not very familiar with. I am a data scientist and still learning how to implement some JS and jQuery in my Shiny application to create a better user experience and more efficient applications.
When you are using the app, you can see that when we add a visualization or remove a visualization, none of the existing visualizations are being re-drawn. This is because I decided against the renderUI function which would have re-created every single visualization after adding or removing them.
Instead, we will be using JS and jQuery to remove and add HTML in the browser without calling the Shiny server too often.
div( class = "class_a", div( class = "col-md-6", div( class = "panel panel-default", div( class = "panel-heading clearfix", tags$h2("Visualization 1", class = "pull-left panel-title"), div( class = "pull-right", shiny::actionButton( inputId = "a", label = "", class = "btn-danger delete", icon = shiny::icon("minus") ) ) ), div( class = "panel-body", plotly::plot_ly(mtcars, x = ~mpg, y = ~wt) ) ) ) )
The class in the first div results in removing the entire visualization and the id inside the actionButton
is the identifier which div should be removed when clicked. Let’s code the logic for that so it will make more sense.
JavaScript + jQuery for Removing Visualizations
// remove plot and add action link $(document).on("click", ".delete", function() { var clicked_id = $(this).attr('id'); var header_h2 = $("#" + clicked_id).parent().parent().text().trim(); $(".class_" + clicked_id).remove(); var html = '<div class="added_' + clicked_id + '"><a id="' + clicked_id + '" href="#" class="action-button">' + header_h2 + '</a>' if ($( "[class^='added_']" ).length) { last_class_added = document.querySelectorAll("[class^='added_']"); added_class = last_class_added[last_class_added.length - 1].className; $(html).insertAfter($("." + added_class)); } else { last_class_added = document.querySelectorAll("[class^='class_']"); added_class = last_class_added[last_class_added.length - 1].className; $(html).insertAfter($("." + added_class)); } })
When we click the delete button, we are going to get the clicked id from the action button which has to be unique with $(this).attr('id')
. In the case above it would be a
. We then grab the header of the visualization, in this case, it is Visualization 1
with$("#" + clicked_id).parent().parent().text().trim()
. We then get the class inside the first div which is class_a
and then remove the entire div with $(".class_" + clicked_id).remove()
. This removes the entire visualization since it lies within that div.
When you look at the final dashboard, you realize that after deleting a visualization, an action link appears in the sidebar. When the action link is clicked, the visualization pops back into the dashboard. So, we have to create the action link next.
shiny::actionLink( inputId = "a", label = "Visualization 1" )
The HTML looks like that: <a id="a" href="#" class="action-button">Visualization 1</a>
However, we have to add a class to it and also a name in order to choose the correct visualization when adding it back to the dashboard.
'<div class="added_' + clicked_id + '"><a id="' + clicked_id + '" href="#" class="action-button">' + header_h2 + '</a>'
The action button link is like this. We are creating an added_a
class and an id equal to a
and add the Visualization 1
header to it.
Afterward, we have to specify where we would like to put the action link. When there is no existing action link, we would like to put the link after the first visualization and when there already exists a link, all subsequent action links will be placed after the last action link on the page with the code below.
if ($( "[class^='added_']" ).length) { last_class_added = document.querySelectorAll("[class^='added_']"); added_class = last_class_added[last_class_added.length - 1].className; $(html).insertAfter($("." + added_class)); } else { last_class_added = document.querySelectorAll("[class^='class_']"); added_class = last_class_added[last_class_added.length - 1].className; $(html).insertAfter($("." + added_class)); }
The code above basically says that when there is no action link, then use the else statement and place it after the last visualization and if there already exists one then place it after the last action link.
That was it for the first tutorial on how I created the application. Please let me know if you have any questions in the comments below. Thank you for the read and look out for the second part!
R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.