General Note

This example is part of a larger set of examples of using Google Analytics with R:

Overview

This example compares two metrics at a page-by-page level to identify high/low performers. This is set to pull Entrances and Bounce Rate and compare them, but additional data is actually included in the query so that the x_dim and y_dim values in the first bit of code below can be updated to compare other metrics. The Shiny app for this example simply allows the user to choose which metrics to view.

# Load the necessary libraries. 
if (!require("pacman")) install.packages("pacman")
pacman::p_load(googleAnalyticsR,  # How we actually get the Google Analytics data
               tidyverse,         # Includes dplyr, ggplot2, and others; very key!
               plotly,            # For interactive charts
               scales)            # Useful for some number formatting in the visualizations

# Authorize GA. Depending on if you've done this already and a .ga-httr-oauth file has
# been saved or not, this may pop you over to a browser to authenticate.
ga_auth(token = ".httr-oauth")

# Set the view ID and the date range. 
view_id <- Sys.getenv("GA_VIEW_ID")
start_date <- Sys.Date() - 31        # 30 days back from yesterday
end_date <- Sys.Date() - 1           # Yesterday

# Set the number of pages to analyze
num_pages <- 100

# You likely won't want to do this for the *entire* site. If you do, then simply enter
# ".*" for this value. Otherwise, enter regEx that filters to the pages of interest (you
# can experiment/test your regEx by entering it in the Pages report in the web interface
# for GA).
filter_regex <- "/blog/.+"

# Set the metrics to use for the x and y dimensions. These will be the "nice label"
# dimensions. We'll look up if/how they need to be calculated.
x_dim <- "Entrances"
y_dim <- "Bounce Rate"

# Set whether the two axes should use a log scale or a standard scale. Set to "log10"
# to use a log scale and "continuous" to use a (typical) linear scale.
x_scale <- "log10"
y_scale <- "linear"

# Create a date frame that we'll use to look up the various calculations for different
# metrics.
calcs_df <- data.frame(metric = c("Entrances", "Bounces", "Unique Pageviews", "Pageviews", "Exits", "Bounce Rate", "Exit Rate"),
                       calculation = c("entrances", "bounces", "uniquePageviews", "pageviews", "exits", "bounces/entrances", "exits/pageviews"),
                       metric_format = c("integer", "integer", "integer", "integer", "integer", "percentage", "percentage"),
                       stringsAsFactors = FALSE)

# Grab the format for x and y from the data frame
x_format <- filter(calcs_df, metric == x_dim) %>% select(metric_format) %>% as.character()
y_format <- filter(calcs_df, metric == y_dim) %>% select(metric_format) %>% as.character()

# Define the base theme for visualizations
theme_base <- theme_light() +
  theme(panel.grid = element_blank(),
        panel.border = element_blank(),
        axis.ticks = element_blank(),
        axis.line = element_line(colour = "gray70"),
        strip.background = element_rect(fill = "white", color = NA),
        strip.text = element_text(colour = "gray20", face = "bold"))

Get the Data

Pull the data and do a bit of wrangling of it.

# Create a dimension filter object. See ?dim_filter() for details. 
page_filter_object <- dim_filter("pagePath", 
                                   operator = "REGEXP",
                                   expressions = filter_regex)

# Now, put that filter object into a filter clause. The "operator" argument is moot -- it 
# can be AND or OR...but you have to have it be something, even though it doesn't do anything
# when there is only a single filter object.
page_filter <- filter_clause_ga4(list(page_filter_object),
                                          operator = "AND")

# Pull the pages data. We'll go ahead and pull a bunch of metrics so that
# we won't have to re-pull if we change x_dim and y_dim.
data_pages <- google_analytics(viewId = view_id,
                              date_range = c(start_date, end_date),
                              metrics = c("entrances", "bounces", "pageviews", 
                                          "uniquePageviews", "exits"),
                              dimensions = c("deviceCategory", "channelGrouping", "pagePath"),
                              dim_filters = page_filter,
                              anti_sample = TRUE)

# Summarize the data using the metrics of interest -- do this overall and then 
# do this broken out by deviceCategory and Channel.

# Get the formulas
calc_details_x <- calcs_df %>%  filter(metric == x_dim)
formula_x <- calc_details_x$calculation

calc_details_y <- calcs_df %>%  filter(metric == y_dim)
formula_y <- calc_details_y$calculation

# First, we're going to roll everything up (removing deviceCategory and channelGrouping), 
# and then we're going to calculate our x and y dimensions. This last step requires NSE,
# which, I realize, has moved on beyond the "_" notation, but I'll be damned if I could get
# that to work.
data_overall <- data_pages %>%  group_by(pagePath) %>% 
  summarise(entrances = sum(entrances), bounces = sum(bounces), pageviews = sum(pageviews),
            uniquePageviews = sum(uniquePageviews), exits = sum(exits)) %>% 
  mutate_(x = formula_x,
          y = formula_y)

# Get the top pages by the x-axis value. We'll use this for our overall plot
top_pages <- data_overall %>% 
  arrange(-x) %>% 
  top_n(num_pages, x) %>% 
  select(pagePath, x, y)

# Calculate the correlation coefficient (r) and the coefficient of determination (r^2)
r <- cor(top_pages$x, top_pages$y)
r_squared <- r^2

# Join this back to the original data and then repeat the calculating. This will give us the
# top num_pages overall...but broken out by device category and channel
top_pages_by_devicecat_channel <- top_pages %>% 
  select(pagePath) %>% 
  left_join(data_pages) %>% 
  mutate_(x = formula_x,  y = formula_y) %>% 
  select(deviceCategory, channelGrouping, pagePath, x, y)

Visualize the Results

We’ll do one visualization of the traffic overall and another visualization that is faceted.

The correlation coefficient overall is: 0.0842237.

The coefficient of determination – which is more commonly referred to as R2 – is: 0.0070936.

# Set up the x and y scales. These vary based on the settings in the initial chunk
format_x <- if(x_format == "integer"){comma} else {percent_format(accuracy = 1)}
format_y <- if(y_format == "integer"){comma} else {percent_format(accuracy = 1)}

if(x_scale == "linear"){
  scale_x <- scale_x_continuous(labels = format_x)
  x_vline <- max(top_pages$x)/2
} else {
  scale_x <- scale_x_log10(labels = format_x)
  x_vline <- max(top_pages$x) %>% sqrt()
}

if(y_scale == "linear"){
  scale_y <- scale_y_continuous(labels = format_y)
  y_hline <- max(top_pages$y)/2
} else {
  scale_y <- scale_y_log10(labels = format_y)
  y_hline <- max(top_pages$y) %>% sqrt()
}

gg <- ggplot(top_pages, mapping = aes(x = x, y = y, text = pagePath)) +
  scale_x +
  geom_vline(xintercept = x_vline, colour = "gray90") +   # vertical line quadrant divider
  scale_y +
  geom_hline(yintercept = y_hline, colour = "gray90") +      # horizontal line quadrant divider
  geom_point(colour = "steelblue", alpha = 0.8) +
  labs(title = paste("Page Analysis: Top", num_pages, "Pages by Total", x_dim, "-", start_date, "to", end_date),
       x = x_dim, y = y_dim) +
  theme_base

ggplotly(gg)
## Same Visualization, but broken down by device category and channel grouping

gg_facets <- ggplot(top_pages_by_devicecat_channel, mapping = aes(x = x, y = y, text = pagePath)) +
  scale_x +
  geom_vline(xintercept = x_vline, colour = "gray90") +   # vertical line quadrant divider
  scale_y +
  geom_hline(yintercept = y_hline, colour = "gray90") +      # horizontal line quadrant divider
  geom_point(colour = "steelblue", alpha = 0.8) +
  labs(title = paste("Page Analysis: Top", num_pages, "Pages by Total", x_dim, "-", start_date, "to", end_date),
       x = x_dim, y = y_dim) +
  facet_grid(channelGrouping ~ deviceCategory, switch = "y") +
  theme_base +
  theme(panel.border = element_rect(colour = "gray50", fill = NA))

ggplotly(gg_facets)