In jdeboer/RGA-1: A Google Analytics API Client

knitr::opts_chunk$set(
    comment = "#>",
    collapse = TRUE,
    error = FALSE,
    tidy = FALSE
)

To access the Reporting API Google Analytics RGA package provides the following functions:

get_ga() - returns Analytics data for a view (profile);
get_mcf() - returns Analytics Multi-Channel Funnels data for a view (profile);
get_realtime() - returns real time data for a view (profile).

All of this functions return a data frame including the data for a view (profile).

The following parameters are available for queries to the API reports:

profile.id - profile (view) ID. Can be obtained using the list_profiles() or via the web interface Google Analytics;
start.date - the start date for collecting data formatted as YYYY-MM-DD. Also values such as "today", "yesterday", "ndaysAgo" are allowed, where n is the number of days;
end.date - the end date for collecting data formatted as YYYY-MM-DD. Also values such as "today", "yesterday", "ndaysAgo" are allowed, where n is the number of days;
metrics - comma-separated list of values of metrics (metrics), for example, "ga:sessions,ga:bounces". The number of metrics cannot exceed 10 indicators for a single request;
dimensions - comma-separated list of values of measurements (dimensions), for example, "ga:browser,ga:city". The number of dimensions cannot exceed 7 measurements at a single request;
sort - comma-separated list of metrics (metrics) and measurements (dimensions) which determine the order and direction of sorting data. The reverse sort order is defined by “-“ before the relevant metric;
filters - comma-separated list of metric (metrics) and measurement (dimensions) filters that are imposed when data is selected;
segment - segments that are used when retrieving data. Can be obtained using the list_segments or via the web interface Google Analytics;
sampling.level - the desired sampling level. Allowed values: "default", "faster", "higher_precision";
start.index - index of the first returned result (line number);
max.results - maximum number of fields (rows) of the returned results;
token - object of class Token2.0 which contains data with a valid authorization data. Can be obtained using the authorize() function;

The following arguments: profile.id, start.date, end.date and metrics are required for get_ga() and get_mcf() (get_realtime() requires only profile.id and metrics). Notice that all arguments must be character strings of unit length. The exception is profile.id which can be both character string and numeric.

This is an example of data obtaining for the last 30 days:

ga_data <- get_ga(profile.id = XXXXXXXX, start.date = "30daysAgo", end.date = "yesterday",
                  metrics = "ga:users,ga:sessions,ga:pageviews")

Sometimes it is necessary to obtain the data for the entire monitoring period through service Google Analytics. For these purposes the package RGA provides the function firstdate() which takes a charming profile ID (submission) as an argument:

first_date <- firstdate(profile.id = XXXXXXXX)

Now we can use the variable first_date as the argument start.date when we call the get_ga() function:

ga_data <- get_ga(profile.id = XXXXXXXX, start.date = first_date, end.date = "yesterday",
                  metrics = "ga:users,ga:sessions,ga:pageviews")

Sampled data

If at least one of the following thresholds is met Analytics samples data accordingly:

1,000,000 maximum unique dimension combinations for any type of query;
500,000 maximum sessions for special queries where the data is not already stored.

In order to avoid this, you can split the query into multiple small queries (day-by-day). You can get this day-by-day data by using the following code:

dates <- seq(as.Date("2012-01-01"), as.Date("2012-02-01"), by = "days")
ga_data <- do.call(rbind, lapply(dates, function(date) {
    get_ga(profile.id = XXXXXXXX, start.date = date, end.date = date,
    metrics = "ga:sessions", dimensions = "ga:keyword",
    filter = "ga:medium==organic;ga:keyword!=(not provided);ga:keyword!=(not set)")
}))
ga_data <- aggregate(. ~ keyword, FUN = sum, data = ga_data)