How to Build Interactive Dashboard in R shiny for Cryptocurrency Analysis
Introduction
Oxford Language Dictionary defines bitcoin as “a type of digital currency in which a record of transactions is maintained and new units of currency are generated by the computational solution of mathematical problems, and which operates independently of a central bank." It was first introduced in 2008 as a decentralized currency by Satoshi Nakamoto. Bitcoin (BTC) is the world's most popular cryptocurrency, inspiring the development of many other cryptocurrencies.
In this step-by-step tutorial, we will learn how to build an interactive dashboard in R shiny to track the historic prices of bitcoin. R shiny app is a great analytical tool that can be used for cryptocurrencies and stocks.
Libraries used
I used the following libraries for my app, but you can modify along the way when creating your own version:
library(shiny) library(shinythemes) library(lubridate) library(dygraphs) library(xts) library(tidyverse)
Data
The dataset for this project was collected from Kaggle.com. It consists of 2747 observations, 7 variables and covers the period from September 17th of 2014 to March 25th of 2022. It doesn’t have any missing values, so there is no need for cleaning. You can find this dataset by following this link.
Use read.csv() function to load the data from the csv file into data frame:
bitcoin <-read.csv(file = 'BTC-USD.csv', stringsAsFactors = F)
Make sure to convert the “Date” column to year/month/day format:
bitcoin$Date <- ymd(bitcoin$Date)
App Design
R shiny package makes it easy to build interactive web applications with R. You will not need in-depth experience with web development, nor will you need to know html, css or javascript languages.
The structure of the shiny app
R shiny app consists of three parts: ui (user interface) object, server and shinyApp functions. In this tutorial, we will use the single file layout meaning that our app will contain all three components in one single app.R file.
UI
The user interface (UI) object is responsible for the layout and appearance of the application. It starts with fluidPage function as demonstrated below:
ui <- fluidPage(
theme = shinytheme("cosmo"),
I used the “cosmo” theme for my app, but there are other options you can explore for your app's overall appearance. Find out more by following this link.
Our sidebar panel which appears on the left side of the application contains the following elements:
- Title of the application
- 5 variables (open, high, low, close, volume)
- Date that ranges from September of 2014 to March of 2022
- The box with the log scale option for price variables only
This can be achieved by using the codes below:
# App title ----
titlePanel(strong("Bitcoin Dashboard")),
sidebarLayout(
sidebarPanel(
h4(strong("Bitcoin closing prices 2014-2022")),
br(),
selectInput('selectOutput',
'Select output',
choices = colnames(bitcoin)[c(2,3,4,5,7)]),
dateRangeInput('selectDate',
'Select date',
start = min(bitcoin$Date),
end = max(bitcoin$Date)),
br(),
br(),
checkboxInput("logInput", "Plot y axis on log scale", value = FALSE),
),
dygraphOutput() function will help us to create the interactive time series graph to visualize the historical prices and the trading volume of bitcoin:
mainPanel(dygraphOutput("priceGraph", width = "100%", height = "800px"))
)
)
Server
This is where the magic happens. Server function makes an application reactive to what you defined in the user interface. It typically starts with the code below:
server <- function(input, output) {
Based on the selected date range, the application will visualize the time series interactive graph for open, high, low, close prices along with the trading volume from our dataframe. In other words, our application will be responsive and change the display according to the user input. Please, note the code for converting the variables into log scale will not work on volume output. This was done on purpose.
getData <- reactive({
# get inputs
selectedOutput <- input$selectOutput
startDate <- input$selectDate[1]
endDate <- input$selectDate[2]
# filter data
data <- bitcoin %>%
select(Date, selectedOutput) %>%
filter(Date >= startDate & Date <= endDate)
# formatting in case of market cap
if(selectedOutput == "Volume"){
data["Volume"] <- lapply(data["Volume"], FUN = function(x) x/1000000000)
}
# converting to logscale for bitcoin price
if(input$logInput == TRUE & input$selectOutput != "Volume"){
data[selectedOutput] = log(data[selectedOutput])
}
data
})
output$priceGraph <- renderDygraph({
data <- getData()
time_series <- xts(data, order.by = data$Date)
dygraph(time_series)
})
}
shinyApp
This is the final step in our web development. This line of code allows us to run the application by binding user interface and server parts.
shinyApp(ui = ui, server = server)
Congratulations! You have just developed your application and it is ready to run. I hope this tutorial was helpful. Here is the complete code for the bitcoin tracker app:
About Author
