Trending Google Analytics sessions

Example 1: Trending Google Analytics sessions

We first load the libraries we will need later, authenticate with Google, and set the View Id we want to pull data from.

library(googleAnalyticsR)
library(knitr)
ga_auth()

## replace with your own Google Analytics View Id
view_id <- 81416156

This is the actual API call. We take advantage of Google Analytics v4’s pivot function, to do some data transformation for us:

## get data, use pivot to get columns of medium sessions
trend_data <- google_analytics_4(view_id,
                                 date_range = c(Sys.Date() - 400, Sys.Date()),
                                 dimensions = c("date"),
                                 metrics = "sessions",
                                 pivots = pivot_ga4("medium","sessions"),
                                 max = -1)

## Note the medium columns created by the pivot
kable(head(trend_data))

Now we have the data, we transform it a little to a format suitable for the plot functions:

## format the names
names(trend_data)

## [1] "date"                     "sessions"                
## [3] "medium.referral.sessions" "medium..none..sessions"  
## [5] "medium.social.sessions"

names(trend_data) <- c("Date","Total","Referral","Direct","Social")

## check the data R classes
## This time all look good - Date objects for date and numeric for the metrics
str(trend_data)

## 'data.frame':    401 obs. of  5 variables:
##  $ Date    : Date, format: "2015-08-07" "2015-08-08" ...
##  $ Total   : num  76 37 48 192 81 68 64 87 35 32 ...
##  $ Referral: num  19 4 11 73 18 18 20 33 12 11 ...
##  $ Direct  : num  12 10 8 59 20 7 9 15 6 5 ...
##  $ Social  : num  0 0 0 0 0 0 0 0 0 0 ...
##  - attr(*, "totals")=List of 1
##   ..$ :List of 1
##   .. ..$ sessions: chr "34563"
##  - attr(*, "minimums")=List of 1
##   ..$ :List of 1
##   .. ..$ sessions: chr "15"
##  - attr(*, "maximums")=List of 1
##   ..$ :List of 1
##   .. ..$ sessions: chr "640"
##  - attr(*, "rowCount")= int 401

Looks good to try some plots:

## make a time-series object
trend_ts <- ts(trend_data[-1], frequency = 7)
plot(Total ~ Date, data = trend_data, type = "l")
axis(1, trend_data$Date, format(trend_data$Date, "%b"))

library(ggplot2)
library(reshape2)

## 
## Attaching package: 'reshape2'

## The following object is masked from 'package:tidyr':
## 
##     smiths

## change data into 'long' format
trend_long <- melt(trend_data, id = "Date")
head(trend_long)

##         Date variable value
## 1 2015-08-07    Total    76
## 2 2015-08-08    Total    37
## 3 2015-08-09    Total    48
## 4 2015-08-10    Total   192
## 5 2015-08-11    Total    81
## 6 2015-08-12    Total    68

## build up the plot
gg <- ggplot(trend_long, aes(x = Date, y = value, group = variable)) + theme_minimal()
gg <- gg + geom_line(aes(colour = variable))
gg

library(highcharter)
library(reshape2)

## change data into 'long' format
trend_long <- melt(trend_data, id = "Date")
hchart(trend_long, type = "line", x = Date, y = value, group = variable)