I’m increasingly reluctant to plug Google services because of the way I’ve started to personally frame their actions as I (cynically) attempt to understand the sorts of process the business appears to use when developing, running and retiring services (if it’s not selling you as audience, Google-as-extractive-industry needs you to be providing it with training signals and data resource to feed its own data-collection tools and algorithms. If there’s no sales, training or data provision to be had, you’re surplus to requirements.). On the other hand, the familiarity of Google can make it an easier sell when trying to persuade people to try out a new application.
But for what it’s worth, here’s a new Google service for helping you get your data onto a map in a relatively straightforward way – Google Maps Engine Lite (launch announcement). (I’m guessing that this is a either a ratcheting sales pitch for the commercial Google Maps Engine (having tried the Lite service out, and decide you may want to make “formal” use of it, you read the Terms and Conditions/license agreement and find you legally can’t…), a way to foster WOM pressure on organisations to adopt Maps Engine (cf. bring-your-own-device – workers or customers tell an organisation “I can do this myself at home why can’t I/you do it here?”), or a way of getting more structured geo-data into Google’s tar pit.)
As a sample data set (the linked to example data set in the help notes is a broken link (missing “http://” relativises it… <- pile of crap WordPress trying to linkify something that isn't a link how do I disable it? Or do I just say: “missing h t t p colon double slash”?), I downloaded (as bp.csv) a data grab I'd made on OpenCorporates of companies co-directed by directors of a sample set of BP companies registered in the UK that includes registered addresses:
To simplify the data (and because I don’t want to give Google more data than I have to, to create a map of companies by registered address), I just grabbed the company name and registration address columns using this handy CSVfilter commandline tool: cat bp.csv | csvfilter -f 0,4 > testdata.csv
(See also: CSVkit)
The 0,4 refer the the 0th and 4th columns in the original data file to give me a datafile that looks like this:
The fact that lots of companies have the same registered address will let me test how the maps engine handles overlaid markers…
The import is then easy enough…
Erm…okay… So we’re limited on the number of markers in each layer, in the “free” plan at least…
(I guess I could try to split my file into 100 row segments and then upload each of them individually. On a Linux/Unix/Mac terminal command line, the command split -l 100 testdata.csv will split the testdata.csv file into separate files 100 lines long (except the last chunk, probably…;-). The filenames are automatically generated by default – inspect the directory listing of the folder you run the command in and sort by most recent date to see the new files that are created.)
We then get walked through identifying columns for locations (more training data for the algos):
and then marker labels:
And then let Google do the geocoding…
Erm… buggritt; does it only work in the US?
The word I use for this is craptastic.
Ho hum, I’ve better things to do than faff around with this any more today…