Hi all,
I've been trying to get the scatter map working, as it's exactly what I need for some important funding reports (that I had already promised that I could do with the wonderful new Analytics!).
However, I have discovered that there is known issue in the Australia & New Zealand region, which means that some post code locations show up incorrectly. TN has an open ticket with Sisense, but no ETA on when that is likely to be resolved.
Some postcodes are showing as being in quite different areas (same city, but very different suburbs/areas). As some of these are our some of biggest selling postcodes, the impact on the map is quite significant and it really skews what we need to show in the reports.
I've tried a number of different orders of parameters in the Location section, but they all return some inaccuracies (best results seem to come from having Country, State, City, Postcode in that order, in the Location section).
Has anyone noticed this same issue and found a good workaround for using scatter maps?
Thanks!
Alison 6281.Scatter map example for TOMATO forum.docx
I've also attached an example :)
Hi Alison, When I was working with Tessitura my best results were with Postcode, State, and Country. City is a little too random and gives me Elasticube errors. I've got a bunch of postcodes that are not mapped for Longitude and Latitude. On the Scatter Map the appear in the most central place in the smallest other location ID
eg: 1300 is a Sydney GPO Box postcode that is unmapped. The orders also have State=NSW and Country=Australia so the scatter map puts it in the middle of NSW
That might be why you are get postcodes dumped in the red square - the most central region of the next location.
That's not really helpful I know but as our dodgy postcodes weren't too big a sample I filtered them out.
Thanks Heath! Your mention of the GPO Box specifically being wrong, made me realise that I think it may only be our PO Box postcodes that are incorrect (initial checks are looking that way so far). Although I can't really change the address to a physical one on all of these records though, which would probably be the only way to resolve it...
How did you know that the postcodes you have were not mapped for longitude and latitude? Is there somewhere you can check?
Thanks
From a help ticket answered by many but this is from the amazing Melissa Champ
The fields from the LOCATION area of the Scatter Map edit screen are used when determining a location’s longitude and latitude on the map. The dashboard filter DOES limit the scope of the Post Codes to post codes in the data. However, the Scatter Map widget then passes just the list of post codes if that’s all that’s in the LOCATION area, to the MapBox resources for longitude and latitude. Therefore if a post code is in the data, all the Scatter Map is doing is taking the post code and trying to determine where that is in the world, without a Country. Therefore the Scatter Map widget should be configured per the documentation to include all the available, helpful location information that will help the Scatter Map to determine a location's longitude and latitude:
Thanks! Scatter maps are obviously a little more unreliable than I'd hoped! I did learn quickly, that you have to include Country, else the postcodes seemed to match all over the world.
I think I will have to make do with using maps only when they are zoomed out a fair way, so it's not quite so obvious of specific locations, and will try to filter out the majority of the PO Box addresses when I can. Hopefully the known issue will be fixed one day soon...