3 ms·
It's not really so simple, binning is a really useful technique to display data, especially large amounts of messy data. Just putting all the points on the map
by cwmma 9y ago
It's not really so simple, binning is a really useful technique to display data, especially large amounts of messy data. Just putting all the points on the map only works if you have a low enough number incidents in a compact enough area that the individual points make sense. But if you are not careful your points might actually be a population density map and be masking higher relative occurrences of something in certain areas.
The key thing is that if your bins (zip codes) don't align well with the underlining data (municipalities) you can mask patterns. In this case they basically (to borrow a phrase from gerymandering) cracked the high led district across a bunch of zip codes.
This is not actually an issue with zip codes nationwide, in MA there is only one zip code there crosses municipal boundaries so if you were doing the analysis in Massachusetts you wouldn't have had this issue.
- will_pseudonym 9y agoYou can easily bin spatial coordinates, too.
- praseodym 9y agoSure, but it still requires some more skill than say, creating an Excel PivotTable and drilling down on the ZIP column.
- phonon 9y agoMake a Per Capita Heat Map. You can chart it in 5 minutes using Google Fusion Tables[1] or Excel with Power Map[2]. Both will even geocode addresses for you! [1] https://support.google.com/fusiontables/answer/1012281?hl=en https://support.google.com/fusiontables/answer/1012281?hl=en [2] https://support.office.com/en-us/article/Get-started-with-Power-Map-88a28df6-8258-40aa-b5cc-577873fb0f4a https://support.office.com/en-us/article/Get-started-with-Po...
- sn9 9y agoIsn't there also often an issue with anonymizing data? More precise spatial coordinates make this harder than with binning methods.