I think Forbes has a much better visualization with more meaningful numbers: http://www.forbes.com/special-report/2011/migration.html

Agreed. For me, arrows would be better to imply movement. Specifically, arrows going from city with the lightness/darkness (tone?) indicating how many are going where. Circles just don't connect with the data part of my brain very well. It's hard to compare the area of circles to gauge just how much bigger one is than the other.

But forbes probably had a team of people working on theirs for a few afternoons at least.

It is a cool tool, but to me it means that people in the SF bay area use the web for housing; other areas not so much.

That assumption doesn't seem to be supported by the data. What are you basing it on?

I thought the same thing, because of the gigantic circles for people moving out of SF, and gigantic circles for moving into neighboring suburbs.

This sort of graph drives me nuts. It's all fancy and interactive, then the data is foolishly selected to be the absolute raw numbers, so the big cities have big circles and everything else doesn't, and the data is useless as all it does is tell me where the big cities are.

Circle size needs to be percentage of population or something weighted towards that to be meaningful, and also throw in a ratio of incoming to outgoing as well to show who is depopulating and who is repopulating.

It would be cool if there were some way to exclude cities within a certain radius. It seems that most searches (inbound and out) are for nearby cities, and the fact that a lot of people move between Boston and Cambridge isn't particularly informative.

Hm, I think it is actually interesting to see how much either suburban or outer cities are looking to move to the "major city" and vice versa.

I was actually kind of interested to see how many people are trying to move from Miami to Ft. Lauderdale.

It'd be easier on the eyes if you added transitions to the circles when the selection changes. For example, in metro_mover.js:

  line 244: .select('circle').transition().duration(500)
  line 258: .select('circle').transition().duration(1000)

One thing that's interesting is that it actually doesn't tell you if more people are moving in or out, despite the title. The size of SF doesn't change when I click inbound/outbound.

I don't believe the amount of people going in or out of the city is represented by the size of the target city.

You're right. The size of the selected city is fixed, just to be visually prominent, while the size of the red/blue related cities is determined by how many searches are happening to/from there.

Does this mean TONS of people are trying to move out of San Francisco?

For those trying to interpret this:

The city highlighted in black is the reference city.

Blue circles indicate cities where people are looking for housing in the reference city (i.e. they are maybe thinking of moving into the reference city)

Red circles indicate cities where people in the reference city are looking for housing (i.e. they are maybe thinking of moving away from the reference city to the cities with red circles)

So for San Francisco, it is not surprising that many people are looking to move out of the city into nearby cities (like Oakland) and vice versa. The same is true for New York.

More interestingly, a number of people in Chicago, Houston and DC are looking for housing in San Francisco, but few people in San Francisco are looking for housing in those cities.

One weakness of the graphic is that it makes it difficult to tell the net effect of inbound and outbound housing searches. Ideally, there would be a third option next to "Inbound Searches" and "Outbound Searches," that showed a net effect for each city using both red and blue circles.

Also, different reference cities are not comparable using the graphic. For instance, selecting SF as the reference city shows moderate inbound and outbound searches from NYC. But select NYC as the reference city and you'll see that SF doesn't make the top 10 cities for outbound or inbound searches and therefore isn't even shown. As a result, you can tell that there are more searches from NY to LA than to SF, but it is difficult to tell how great that difference is.

Thanks for taking the time to explain the visualization (I didn't build this, but helped with some of the bugfixes). We're super interested in looking at this kind of data, and it's not quite the easiest to explain. There are a lot of small cases that would disappear if we were only looking at the net inbound/outbound, but adding it as a third mode is a good idea.

For reference, we also have a companion blog post to this visualization here: http://insights.truliablog.com/2012/02/where-are-house-hunte...

And a longer post on an index based around that net inbound/outbound searching, what we're calling the Metro Movers Index: http://insights.truliablog.com/2011/11/trulia-metro-movers-f...

It will be cool if I can compare several cities.