tag:blogger.com,1999:blog-47896653854399647.post2573595774111622940..comments2024-03-19T08:43:55.574+00:00Comments on Google Earth Design: 3 Crime Maps: Point collationRich Treveshttp://www.blogger.com/profile/00074610839485450953noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-47896653854399647.post-58257854026875365342011-08-01T18:25:43.394+01:002011-08-01T18:25:43.394+01:00Katy,
Doh! I've done that before too....
th...Katy,<br /><br />Doh! I've done that before too....<br /><br />thanks<br /><br />RichRich Treveshttps://www.blogger.com/profile/00074610839485450953noreply@blogger.comtag:blogger.com,1999:blog-47896653854399647.post-51185200467114341832011-08-01T11:54:19.917+01:002011-08-01T11:54:19.917+01:00Interesting links, in the post and in Steven's...Interesting links, in the post and in Steven's comment - thank you.<br /><br />One comment, though - it's choropleth (not chloro-) :)Katynoreply@blogger.comtag:blogger.com,1999:blog-47896653854399647.post-16141294996916682712011-07-28T10:02:06.175+01:002011-07-28T10:02:06.175+01:00Steven,
had a look at your links. ukcrimestats i...Steven,<br /><br />had a look at your links. ukcrimestats is 'vanilla' google maps: they don't do any collation.<br /><br />re extents: yes, yes, yes! Glad someone else thinks this too. The post could have easily expanded into discussion of this but I didn't have all day.<br /><br />re cms.geo.me, my gut feeling is that doesn't deal with the issues: The lowest level of clustering doesn't have symbol size = number in cluster. I'm also suspicious of the symbols obscuring each other. <br /><br />However, I REALLY want to see how user tests go on these sorts of visualizations because they may have positive aspects that haven't occurred to me.<br /><br />Finally, I like clustering as a term better than collation.Rich Treveshttps://www.blogger.com/profile/00074610839485450953noreply@blogger.comtag:blogger.com,1999:blog-47896653854399647.post-59853281164701322972011-07-27T14:22:51.500+01:002011-07-27T14:22:51.500+01:00Have a look at http://www.ukcrimestats.com/ they d...Have a look at http://www.ukcrimestats.com/ they do some pretty impressive stuff with the raw data from police.uk Clustering is still a problem though.<br /><br />Clustering is inherently misleading, if additionally the raw data has already been aggregated then you need to be doubly careful. Ideally the cluster symbols need to incorporate some extents for the cluster so that when you click on the symbol you zoom to the extents of the cluster and see all of the points. Not easy to do dynamically.<br /><br />My mates at geo.me are doing some neat stuff with clusters. This is a rough proof of concept not quite there but worth a look http://cms.geo.me/app/scdcbus<br /><br />Cheers<br /><br />StevenStevenhttps://www.blogger.com/profile/03027329503460961288noreply@blogger.com