Heat Maps for Google Maps – (a.k.a GeoIQ mashup)
October 11th, 2006by Sean Gorman
About the Author: Sean Gorman founded FortiusOne in 2005 to bring location based analytics to the mass market. Sean brings over 10 years of experience at the forefront of the geospatial revolution as a researcher, practitioner, and entrepreneur at FortiusOne. Through both academic and entreprenurial efforts he has been working to make geographic data more accessible to the public since 1997 culminating in the creation of GeoCommons – a crowd-sourced repository of statistical data and social feeds that can be easily mapped, remixed and reused by non-technical users.
Sean has been featured in media such as, Wired, Der Spiegel, ABC, Washington Post, Business 2.0, MSNBC, CBS and CNN. He also holds a PhD. From George Mason University in Public Policy where he was the Provost’s High Potential Scholar and was the recipient of the Fischer Prize. He has published dozens of articles on geographic data sharing and analysis, and authored the book Networks, Complexity and Security: The Role of Public Policy in Critical Infrastructure Protection. Read more from this authorSo it has been a while since we posted, but the rationale was we’d wait till we had a working example of moving past push pins. This week we got our GeoIQ API working with the Google Maps API and have the first set of screen shots to show. One of the things we thought is really missing from web mapping applications, right now, is the ability to do geographic analysis. Even the ability to make basic decisions like – is location “A” better than location “B” is missing. With this first simple idea in mind we’ve built a quick mashup with Google Maps. We took our heat mapping API and integrated it with a split screen Google Maps viewer. That way you can look at two locations at the same time and compare them.
We wanted a fun data set to play around with and thought traffic congestion/delay would be interesting. The Bureau of Transportation Statistics (BTS) has a cool data set with average traffic delay for all the US highways available, so we threw that in. One of the problems with pushpins or polylines in Google Maps (and others) is there is no way to visualize what are the high value or low value pushpins. In this case, which road has high traffic delay and which roads have low traffic delay. We do this with a heat map (similar to Zillow, Google Adsense, etc.) that can be dynamically refactored as you zoom in/out (see previous post). We added to this heat map tool a concentration index – which gives you a score of the value (weight) of your pushpins and how closely they are located together. Once you have the score you can see if location “A” is better than location “B”. In this case is traffic delay more concentrated in location “A” or location “B”
A comparison of the concentration of traffic delay in San Francisco and Los Angeles
The GeoIQ API creates a heat map based on an index that measures the amount of traffic delay on the roads and how closely that road delay is located to other delayed roads. The higher the delay and the closer together the roads, the hotter the map and the higher the score. The score ranges between 0 and 1. If all the traffic delay and highways were concentrated at one single location the score would 1 and if there was no traffic delay the score would be 0. In the map above traffic delay for Los Angeles in .26 and for San Francisco it is .15, so if you believe the BTS data traffic, LA is about twice as bad as SF. Lets go east coast – NYC vs. DC.
A comparison of the concentration of traffic delay in New York and Washington DC
According to this score NYC is a little worse than DC. The cool thing about the technology is you can run these comparisons on the fly as you zoom in and out of the map. So – let’s compare two big traffic bottlenecks in DC to see which is worse the I-270 Spur or I-95 Mixing Bowl.
A comparison of the concentration of traffic delay at the I-270 Spur and the I-95 mixing bowl – both in Washington DC
The Spur looks to get the better of the Mixing Bowl. In the app you can do this with any data set or mash up multiple data sets to solve a variety of problems surround location decisions. We’ll have more to come so stay tuned if this looks interesting.
Popularity: 100% [?]
October 11th, 2006 at 7:46 pm
Cool stuff!
October 11th, 2006 at 11:05 pm
Looks great, definitely a big increase in functionality over traditional GEarth mash-ups.
October 12th, 2006 at 11:41 am
Very fun! When will there be a demo up that we can play with? I’d like to try testing some of my own locations.
October 13th, 2006 at 7:58 am
That looks like a really nice idea, but I think your pictures illustrate some of the problems with heatmaps – it’s very hard to convert that colour scale back to the original values. For example, comparing the spur and the mixing bowl, the spur has a lower value (0.16) but it looks like a brighter center point. I’d recommend experimenting with using other ways of representing the same data. For example, you could place points at each measurement location and then make the size of the point proportional to the amount of traffic (aka a bubble chart). This would also remove the uncertainty over whether an area has no traffic, or no measurements.
I’m also not sure about your smoothing algorithm – it would be nice (but difficult) to take into account the actual geography of the roads, traffic density falls off abruptly on the sides of the road, not gradually!
October 13th, 2006 at 9:36 am
Heat maps (map algebra – raster analysis) are one of the spatial tools you can use for geographic analysis of data. It does have its strengths and weaknesses, just like any other analytic. Although I do think we should clear up a little bit what those are.
Sure you could drop proportional symbols on top of the map – it is not terribly hard to do, and we’ll probably throw it in for a later release. You have a nice review of the benefits of proportional symbols, but there is also some drawback that heatmaps do much better. Proportional symbols do not work very well for polyline data, but regardless it would tell you the same thing as the heat map – the ramps connecting the major arteries of the I-270 spur (according to BTS) do not have as much traffic. That said sometimes proportional symbols are a better choice for a map, and that is up to the user to decide. Let’s not forget before today there were zero choices for Google Maps.
What you cannot get with proportional symbols is a score to compare two location. Secondly you cannot not integrate more than one variable to make a location decision. Thirdly they scale terribly – what happen when you try to draw 80,000 bubbles at the national level for the entire highway system. Lastly, you cannot do any of the fancy map algebra analytics like looking at temporal changes.
To answer the last question – you can take into account distance with smoothing algorithm. Spatial mathematics (splining, kriging, kernel density) to make heat maps require that you set a distance threshold. We do this dynamically as you zoom in so
October 13th, 2006 at 9:38 am
you get an increasing about of detail as you go. You can also set it yourself with a slide bar. Since it is a Gaussian distance decay function it will search out for attributed data as far as you tell it to. The result is a surface not a line. If we had geo-referenced data on different delays for different lanes you could see that, but BTS only has one figure for each road. The distance decay fade does not mean there is delay in the medians or on the shoulders it is just the way the mathematics works.
You are of course free to use whatever tool you would like. We are just trying to make more tools available to the public so they pick the right one to solve the problem at hand.
Re-reading this post also makes me realize we are really getting into the academic nitty gritty of analytics, and our whole goal was for users not have to deal with that level of detail to solve a problem. We are happy to explain how we do thing and the mathematics and rigor behind them at any point so feel free to ask.
October 13th, 2006 at 6:09 pm
Heat Maps for Google Maps
"So it has been a while since we posted, but the rationale was we ’d wait till we had a working example of moving past push pins. This week we got our GeoIQ API working with the Google Maps API and have the first set of screen shots to show."
October 13th, 2006 at 8:39 pm
I think this was talked on eUNKNOWN and they talked how it is not 100 percent correct or how it would not be 100 percent correct or somthing like that.
October 14th, 2006 at 12:47 am
Uhhhh yeah… anyone who says the Spur is worse than the Mixing Bowl needs to have their head examined (or logic reprogrammed). Just a thought from someone who actually drives the area.
October 14th, 2006 at 1:59 am
[...] Heat Maps for Google Maps The GeoIQ API creates a heat map based on an index that measures the amount of traffic delay on the roads and how closely that road delay is located to other delayed roads. (tags: googlemaps maps mashup data database visualisation) [...]
October 14th, 2006 at 8:16 am
Another thought that occured to me was the use of a time dimension.
The use of a slider could provide you with the ability to “slide” across a timeperiod (a day, a week, perhaps more). The concentrations across a time scale would be of particular interest to people who are also trying to perform a trend analysis. Just my initial thought.
Interesting approach -> can be applied to a whole gamut of GIS data; pollen, pollution levels, even by grocery stores to evaluate product sale.
October 16th, 2006 at 2:24 pm
Here’s a color-coded / thematic map integrated with Google Maps. Includes rollover / hover text.
http://www.gmapsgis.com/examples/2004election.php
October 16th, 2006 at 11:33 pm
Good post: the only thing better that knowing, is knowing together.
October 17th, 2006 at 3:20 am
[...] Moving Past Push Pins » Blog Archive » Heat Maps for Google Maps – (a.k.a GeoIQ mashup) (tags: cool google Maps Googlemaps mashup heatmap visualization traffic) [...]
October 17th, 2006 at 5:47 pm
[...] FortiusOne is about to release (with an API trial key) their GeoIQ API that works on top of the Google Maps API. GeoIQ should provide the geographic analysis facilities that Google Maps lacks. Their first idea has been to integrate their heat mapping API. In their blog, they provide an example of heat maps of traffic congestion based on the US Bureau of Transportation Statistics. [...]
October 18th, 2006 at 10:59 am
[...] "So it has been a while since we posted, but the rationale was we ’d wait till we had a working example of moving past push pins. This week we got our GeoIQ API working with the Google Maps API and have the first set of screen shots to show."read more | digg story [...]
October 20th, 2006 at 5:33 pm
Awesome! Later, things like road conditions can be modeled more precisely. For example, a highway may be somewhat congested but still move faster than a small road. And the congestion may be only in one direction.
January 14th, 2007 at 6:05 pm
Good post: the only thing better that knowing, is knowing together.
January 27th, 2007 at 6:35 pm
[...] Another cool way to waste time: Heat Maps for Google Maps – (a.k.a GeoIQ mashup) [...]
March 6th, 2007 at 6:26 pm
[...] Original post by unknown [...]
March 6th, 2007 at 9:19 pm
[...] Original post by unknown [...]
August 27th, 2007 at 1:47 pm
October 5th, 2007 at 1:45 pm
There’s one special secret Sale link on Amazon:
http://tinyurl.com/2r7ldr
[b]This is open every Friday and ONLY on Fridays! [/b]
You can find very good discounts here, although some Fridays you can really get
lucky and make off like an Amazon bandit – I´ve seen discounts there as low as 75%
off sticker Price.
October 7th, 2007 at 4:04 am
So zillow has a heat map, but it is a still shot of the market prices. Has anyone ever tried to apply the dimension of time to this, rendering a market price rate of change? For example… market trend over the past month, the past quarter, 6 months, year, 5 years, etc. If anyone knows how this could be done, PLEASE contact me. I am vitally hungry for this information!
October 22nd, 2007 at 2:34 am
I was playing poker the other night with Tarot cards I got a full house and 4 people died — Steven Wright
October 22nd, 2007 at 11:46 am
Hello
Good news guys
I here I with you
http://cialis20mg.org
October 23rd, 2007 at 9:37 am
[...] read more | digg story [...]
October 28th, 2007 at 6:15 am
Hello Men’s
you must trasted
October 31st, 2007 at 3:59 pm
Hello
stgdgedr
restartos
January 20th, 2008 at 2:13 am
hi wats upimon a laptop and hi
April 1st, 2008 at 3:02 pm
Prescription drugs without a prescription and generic medications. Canadian online medicine.
http://canadian-online-medicine.com
September 25th, 2008 at 8:09 pm
site google home green mail ibm
October 26th, 2008 at 3:38 pm
Sounds crazy. I like the way you react to my adventures techniques I have a nice joke. What does a dog get when it finishes obedience school? A pet degree.
April 23rd, 2010 at 11:02 pm
Do you know if there is any published library for generating these heat maps from a KML file? I’d love to be able to do some geo-visualization like this.
June 16th, 2010 at 12:13 pm
Now added to our newly released Google heatmap service…
a PHP driver code example.
Start at: diffent.com/heatmap
Example in JavaScript also but can be driven from any HTTP POST driver.
August 15th, 2010 at 4:50 am
The HeatMapAPI is an API that enables anyone to create their own heat maps. If you wish to use the HeatMapAPI the GeoChalkboard blog has a good tutorial on how to create a heat map with the API and Google Maps.