As part of our research on the geography of search engines, we coded a new R tool, including ideas from my colleague Simon Scheider (Universiteit Utrecht). Google Trends is a great source of “big data” about search behaviour at a large scale, but it limits the number of query terms to 5. Because of the way the Google Trends index is scaled between 100 and 0, many terms fall to 0, and are practically unusable.
The Google Trends comparator overcomes this limitation, allowing an arbitrary number of terms as input for the Google Trends API, and producing a new index scaled between 0 and 1000.
[GitHub repository with code and examples]
For example, for 10 terms Austria, Belgium, Cyprus, Denmark, Estonia, Finland, Germany, Greece, Hungary, and Italy, the script produces this table, where scaled_hits is the Google Trends index, and hits1000 is the same value rescaled between 0 and 1000 for readability. Note that hits is the average of the Google Trends index in the selected time_span.
| term | scaled_hits | hits1000 | collected_at | time_span |
|---|---|---|---|---|
| Austria | 101.9617 | 324.2405 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Belgium | 123.5705 | 392.957 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Cyprus | 74.1176 | 235.6956 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Denmark | 76.9148 | 244.5908 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Estonia | 19.8276 | 63.0522 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Finland | 53.7647 | 170.9729 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Germany | 314.4632 | 1000 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Greece | 151.9191 | 483.1061 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Hungary | 35.1315 | 111.719 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
| Italy | 299.5809 | 952.674 | 2018-12-11 11:22 | 2017-12-17 2018-12-02 |
As always, suggestions and feedback are welcome.
Data Science in Place 