Damien Fleminks 21/11/2020
# rvest for web scraping
# -> install.packages('rvest')
library(rvest)
# tidyverse & gridExtra for data manipulation & visualization
# -> install.packages("tidyverse")
# -> install.packages("gridExtra")
library(tidyverse)
library(gridExtra)
# httr & jsonlite for rest api
# -> install.packages("httr")
# -> install.packages("jsonlite")
require("httr")
require("jsonlite")
# rlist to extend list functions
# -> install.packages("rlist")
library(rlist)Use coinranking.com to get top 50 cryptocurrency coins and retrieve the ticker symbol, coin name & current market cap
crypto.url <- read_html("https://coinranking.com/")
crypto.tickers <- crypto.url %>%
html_nodes(".profile__subtitle") %>%
html_text()
crypto.name <- crypto.url %>%
html_nodes(".profile__link") %>%
html_text()
crypto.market.cap<- crypto.url %>%
html_nodes(".valuta") %>%
html_text()Clean up the strings by removing unused characters and split into separate vectors
crypto.tickers <- str_trim(str_remove_all(crypto.tickers, "\n"))
crypto.name <- str_trim(str_remove_all(crypto.name, "\n"))
crypto.unit.price <- crypto.market.cap[c(TRUE,FALSE)]
crypto.market.cap <- crypto.market.cap[c(FALSE,TRUE)]Convert to correct data types and track which market cap price uses price unit in billion so we can multiply all output into units of million
crypto.isbil <- str_detect(crypto.market.cap,"billion")
crypto.unit.price <- as.numeric(gsub("[^0-9.]", "", crypto.unit.price))
# find billion values and change them into million units
crypto.market.cap <- as.numeric(gsub("[^0-9.]", "", crypto.market.cap))
for(i in 1:length(crypto.isbil)){
if(crypto.isbil[i]){
crypto.market.cap[i] <- crypto.market.cap[i] * 1000
}
}
# Create data frame of scraped and formatted data
crypto.currencies <- data.frame(Cryptocurrency = crypto.name, Ticker = crypto.tickers, Price=crypto.unit.price, market.cap = crypto.market.cap)We now have a cleaned up Data frame with currency data that was taken from coinranking.com. We will use this data to retrieve historical price data from an api.
print(
head(crypto.currencies)
)## Cryptocurrency Ticker Price market.cap
## 1 Bitcoin BTC 18252.0400 338870
## 2 Ethereum ETH 562.2700 63630
## 3 XRP XRP 0.5489 24200
## 4 Tether USD USDT 1.0000 19730
## 5 Litecoin LTC 76.3000 5060
## 6 Polkadot DOT 4.8900 4990
With the data collected from the web we will retrieve historical pricing data from api.tiingo.com API, we will start by by setting up the URL to execute a GET request. An account with API token is needed from api.tiingo.com, I requested a free token which had a limit of 20K requests a day which should be more then enough.
# Setting up base URL for cryptocurrency price history API
api.base <- "https://api.tiingo.com/tiingo/crypto/prices"
# Set currency used for pricing cryptocurrencies
api.currency <- "usd"
# Retrieve history for top 5 cryptocurrencies, using the tickers from our cryptocurrency data frame
api.tickers <- paste(crypto.currencies$Ticker[1:5],"usd",sep="")
api.tickers <- paste(api.tickers, collapse = ',')
# Start date to retrieve prices in YYYY-MM-DD format, set 1 year from today
api.startdate <- as.POSIXlt(Sys.Date())
api.startdate$year <- api.startdate$year-1
api.startdate <- as.Date(api.startdate)
# set the frequency in which you want data resampled
api.resampleFreq = "1day"
# API token for authentication (A free token can be requested at: api.tiingo.com, but you can use mine for now)
api.token = "20b7879009ee4863aef6a4dfd06fb50fb9ea447e"
# Combine variables into GET URL
api.call <- paste(api.base, "?tickers=",api.tickers,"&startDate=",api.startdate,"&resampleFreq=",api.resampleFreq,"&token=",api.token,sep='')Make GET request with the request URL and store response in variable, we will convert the result into JSON and clean up the data and format it into a usuable format for our analysis.
api.prices <- GET(api.call)
# Format response to JSON
api.prices_text <- content(api.prices, "text")
api.prices_json <- fromJSON(api.prices_text, flatten = TRUE)
api.prices_df <- as.data.frame(api.prices_json)
# Create a single data frame with formatted data for easier data visualization and correct data types
returnNewDf <- function(index){
# Get coin and store dataframe
temp.coin <- api.prices_df$baseCurrency[index]
temp.df <- api.prices_df[index,][[4]][[1]]
# Get length and store coin name as new column
temp.rownumber <- length(temp.df[,1])
temp.df$coin <- rep(c(temp.coin),each=length(temp.rownumber))
temp.df$date <- as.Date(temp.df$date)
# To calculate the price value of max in terms of percentage add new column with percentages of close
temp.df$price.percentage <- temp.df$close/max(temp.df$close)
return(temp.df)
}
# Add first data frame, then loop to add the rest
crypto.all <- returnNewDf(1)
for(i in 1:length(api.prices_df[,1])) {
crypto.all <- rbind(
crypto.all,returnNewDf(i)
)
}Now we can access all coins from the same data frame with the different unique coins, looking at the head of 2 examples
unique(crypto.all[c('coin')])## coin
## 1 btc
## 735 eth
## 1102 ltc
## 1469 xrp
print(
head(filter(crypto.all, coin=='btc'))
)## high date low tradesDone close open volume
## 1 7661.697 2019-12-09 7274.841 438772 7342.620 7517.407 65033.53
## 2 7409.128 2019-12-10 7159.498 514511 7225.307 7340.929 61604.03
## 3 7276.874 2019-12-11 7128.879 336152 7210.088 7225.744 40023.15
## 4 7302.802 2019-12-12 7084.571 407312 7198.602 7210.588 57418.48
## 5 7308.903 2019-12-13 7191.353 323726 7258.717 7197.521 38007.89
## 6 7272.773 2019-12-14 7012.930 347467 7067.233 7258.641 38128.94
## volumeNotional coin price.percentage
## 1 477516508 btc 0.3727445
## 2 445108043 btc 0.3667892
## 3 288570449 btc 0.3660166
## 4 413332834 btc 0.3654335
## 5 275888502 btc 0.3684852
## 6 269466107 btc 0.3587647
print(
head(filter(crypto.all, coin=='eth'))
)## high date low tradesDone close open volume
## 1 151.2740 2019-12-09 146.6895 109866 147.4179 150.5599 460701.9
## 2 148.3835 2019-12-10 143.8853 110471 145.6068 147.4286 534537.8
## 3 146.3632 2019-12-11 142.1659 93453 143.4108 145.5938 401814.1
## 4 145.9552 2019-12-12 139.2701 127244 144.8781 143.4336 615179.0
## 5 145.5472 2019-12-13 142.8940 102316 144.7974 144.8458 429064.6
## 6 145.0951 2019-12-14 141.1907 89769 141.8480 144.8032 337352.5
## volumeNotional coin price.percentage
## 1 67915715 eth 0.2390601
## 2 77832336 eth 0.2361231
## 3 57624482 eth 0.2325620
## 4 89125973 eth 0.2349414
## 5 62127431 eth 0.2348105
## 6 47852788 eth 0.2300277
We can analyze the price of the coins over time to see how they have performed over the past year.
# Plot a line graph for all coin prices
cPrice1 <- ggplot(filter(crypto.all, coin=='btc'), aes(x=date,y=close)) + geom_line() + ggtitle("Bitcoin")
cPrice2 <- ggplot(filter(crypto.all, coin=='eth'), aes(x=date,y=close)) + geom_line() + ggtitle("Ethereum")
cPrice3 <- ggplot(filter(crypto.all, coin=='link'), aes(x=date,y=close)) + geom_line() + ggtitle("Chainlink")
cPrice4 <- ggplot(filter(crypto.all, coin=='xrp'), aes(x=date,y=close)) + geom_line() + ggtitle("XRP")
grid.arrange(cPrice1 , cPrice2, cPrice3, cPrice4, nrow = 2)
As we have added an extra column to compare the percentage of the price to the max price we can compare how all coins compared relatively from each other in terms of price fluctation.
# Compare the different coins on the percentage growth
ggplot(crypto.all, aes(x=date,y=price.percentage)) + geom_line(aes(color=coin)) + ggtitle("Price Fluctation") 
From the tradesDone and volumeNotional column we can analyze the number of transitions and total volume in USD traded, we will compare the results and look at the daily trend over the year and the total numbers.
# Compare frequency & volume in USD of trades over the last year
cTradeComp <- ggplot(crypto.all, aes(x=date,y=tradesDone)) + geom_line(aes(color=coin)) + ggtitle("Transactions per day")
cVolComp <- ggplot(crypto.all, aes(x=date,y=volumeNotional)) + geom_line(aes(color=coin)) + ggtitle("Volume in USD per day")
grid.arrange(cTradeComp , cVolComp, nrow = 2)
# Compare total frequency & volume of trade
cTradeTotal <-ggplot(crypto.all) + geom_bar(aes(y=tradesDone, x=coin, fill=factor(coin)), stat="identity") + ggtitle("Total transactions")
cVolTotal <- ggplot(crypto.all) + geom_bar(aes(y=volumeNotional, x=coin, fill=factor(coin)), stat="identity") + ggtitle("Total volume traded in USD")
grid.arrange(cTradeTotal , cVolTotal, nrow = 1)