Addresses on a BI map

Power BI, Tableau and Metabase place a row on a map best when the row has a latitude and a longitude. Their built-in matching works on country, region and city names; a street address needs a geocoder.

New Geocode addresses R, Spark, BI Command line Retail, e-commerce

Add the coordinates before the dashboard

terminalCLI
gridpin batch usa.bin stores.csv stores_geocoded.csv
gridpin batch usa.bin stores.csv stores_geocoded.csv --with h3:7

The output keeps every column and adds gp_lat, gp_lon and gp_match_level. Point the BI tool at the new file. --with h3:7 adds a hexagon cell to colour areas instead of single points.

Keep the weak rows off the map

Filter on gp_match_level: PointAddress is the house, StreetName only the street, Locality only the town. A dashboard of deliveries should not draw a town centre as if it were a customer. gridpin.dq in Python gives the share of a column that geocodes to the house, as a data-quality check before the refresh.