Having never been to the US and having no short-term plans to change that, I find the data itself somewhat ... not so interesting, I guess.

But the way the data is visualized, and the way you can explore it, is awesome. It's like a database you can query just by moving the mouse around. Strike the "like" - that is exactly what it is.

Thank you for the kind words. You know, even if you are not interested in US data, you now have the piece of knowledge in your memory installed there in the most aesthetic and gentle way, I call it 'visually transmitted knowledge' J

How do you explain the last decade of your graph?

The totals look to drop sharply by about 70%, but looking at the data, it was pretty consistent at about 1 million a year.

In general I have a problem with applying smooth curves to data derived from histograms.

The drop at the end is because official website of the Department of Homeland Security has data only till 2013, and it is only four years of data versus ten for all the other points. What is the problem with smooth curves?

neat!

the drop at the end could easily make a reader think that immigration has dropped recently. to avoid giving that misimpression, i'd recommend to multiply the last datapoints by 10/4; equivalently, think of your y-axis units as immigrants per year, rather than immigrants per 10-year bucket.

the problem with smooth curves is that they create data that doesn't exist; like someone who invents a few extra details when telling a story in order to make it more interesting. there's no clear line between editorializing, clarifying, embellishment and fabrication; even a blue dress can look white.

anyway, smooth curves obscure the limits of your data; without them it'd be more clear that we're strictly dealing with decades. for your most interested readers that clarity would be helpful.

I agree that the drop on the graph may give impression of decrease in immigration, so thank you for your idea about scaling. I try to keep data 'as is' and give only new image of it but not new data. Scaling makes much more sense (I used it in visualization of Israel immigration because their statistic office has VERY strange approach to period slots).

Regarding smooth curves, you see, the data is unchanged - you can see exact numbers in tooltips, the visualization is just new look of it, what it's supposed to be )