Tue, 08 Mar 2016

gtrendsR 1.3.3

A very nice new update to the gtrendsR package by Philippe and myself is now avilable via CRAN. I had only blogged about the initial 1.3.0 release, and we have added a whole slew of new features and fixes. Philippe rewrote a lot of the parsing to make it more robust to different encodings, and to add other features. So in no particular order, we can now sub-group by regions more finely, withstand various misfeatures in returned data sets, generally do better on connections, and more --- and also allow for intra-day, daily and weekly queries!

That last part is pretty fun. Here is the code I ran last Saturday to look at the query for Donald Drumpf, a name brought to us via a beautiful John Oliver episode worth watching which ran about nine days ago. So last Saturday, when we were still within the seven day window, I ran

dp <- gtrends("Donald Drumpf", res="7d")
plot(dp) + ggplot2::ggtitle("The Drumpf") + ggplot2::theme(legend.position="none")

which resulted in the following chart

Donald Drump query 

which highlights another nice feature: the ggplot2 object created by the plotting function is returned, so we can locally modify and tune it. Here we set a title and suppress the default legend.

As I had not blogged about the interim bug-fix releases 1.3.1 and 1.3.2, here is the set of NEWS entries for the last three releases:

gtrendsR 1.3.3

  • A ggplot2 object can now be returned for further customization. plot(gtrends("NHL")) + ggtitle("NHL trend") + theme(legend.position="none")

  • Support for hourly and daily data (#67). For example, it is now possible to have hourly data for the last seven days with gtrends("nhl", geo = "CA", res = "7d"). Use ?gtrends for more information about the time resolution supported by the package.

  • Support for categorties (#46). Ex.: gtrends("NHL", geo = "US", cat = "0-20") will search only in the sport category.

  • Some countries (ex: Hong Kong) were missing from the list (#69).

  • Various typos and documentation work.

gtrendsR 1.3.2

  • Added support for sub-countries (#25). Ex.: gtrends("NHL", geo = "CA-QC") will return trends data for Québec province in Canada. The list of supported sub-countries can be obtained via data(countries).

  • Data parsing should work for any data returned by Google Trends (i.e. countries independent).

  • Better support for queries using keywords in different languages (#50, #57). Ex.: gtrends("蘋果", geo = "TW")

  • Now able to specify up to five countries (#53) via gtrends("NHL", geo = c("CA", "US"))

  • Fixing issue #51 allowing UK-based queries via geo = "GB"

gtrendsR 1.3.1

  • Fixing issue #34 where connection verification was not done properly.

  • Now able to use more latin character in query. For example: gtrends("montréal").

  • Can now deal with data returned other than in English language.

Courtesy of CRANberries, there is also a diffstat report for the this release. As always, more detailed information is on the gtrendsR repo where questions, comments etc should go via the issue tickets system.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

Edited 2016-03-08: Corrected code snipped and one grammar instance

/code/gtrendsr | permanent link