Comments on: Symbolizing point data in SpatialKey http://blog.spatialkey.com/2008/09/symbolizing-point-data-in-spatialkey/ Geotemporal visualization: theory + solutions Mon, 19 Apr 2010 18:46:58 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: Javier de la Torre http://blog.spatialkey.com/2008/09/symbolizing-point-data-in-spatialkey/comment-page-1/#comment-397 Javier de la Torre Thu, 02 Apr 2009 07:30:17 +0000 http://blog.spatialkey.com/?p=121#comment-397 Another kind of visualization that I like is "spatially balanced points". The best example is the Panoramio, Youtube and Wikipedia layers in Google Maps. These layers have millions of points behind. This representation is not very interesting for data analysis or find trends, but I like it when your point is to let the user browse the data randomly and get and idea of what data is behind this huge dataset. In the case of Wikipedia, Google can apply some ranking magic and at low zoom levels you get articles about the countries, continents and so on. But on the case of the pictures I think they just pick them randomly. The key is to spatially balance the data so that the user gets interested in further zooming in the map. Again, not a very interesting visualization for data trends findings, but an engaging visualization to explore a dataset. Regarding heatmaps the only problem i see with them is the user having to play with the scale tool to get a nice visualization depending on how his data is distributed. If your data has one particular place with lot of data and the others with little, the default heatmap you get kind of hide the general trend because it only highlight this particular place with lot of data. So instead of having a linear scale sometimes you might want exponential or things like that. Another kind of visualization that I like is “spatially balanced points”. The best example is the Panoramio, Youtube and Wikipedia layers in Google Maps. These layers have millions of points behind. This representation is not very interesting for data analysis or find trends, but I like it when your point is to let the user browse the data randomly and get and idea of what data is behind this huge dataset.

In the case of Wikipedia, Google can apply some ranking magic and at low zoom levels you get articles about the countries, continents and so on. But on the case of the pictures I think they just pick them randomly. The key is to spatially balance the data so that the user gets interested in further zooming in the map.

Again, not a very interesting visualization for data trends findings, but an engaging visualization to explore a dataset.

Regarding heatmaps the only problem i see with them is the user having to play with the scale tool to get a nice visualization depending on how his data is distributed. If your data has one particular place with lot of data and the others with little, the default heatmap you get kind of hide the general trend because it only highlight this particular place with lot of data. So instead of having a linear scale sometimes you might want exponential or things like that.

]]>