I mentioned the ‘old’ Google Charts utility yesterday, which seems to have been succeeded with this system.
There are shinier graphics utilities out there, but the old system, which produces a .png, lends itself outstandingly to use in maps, mostly because the code to generate the image all fits on one line, meaning that you can manage and customize them in Excel and easily mass-produce individualized graphs for each placemark, regardless of number. The .pngs are only generated as needed.
I wasn’t sure if Fusion Tables would accept it, but it does.
Here we use it in a line chart of provincial crime rates. I was pleased with the way it turned out, having been frustrated by Google Charts line charts in the past.
There are some quirks, to put it mildly. The position of a point on the vertical axis reflects its position on a 0-100 scale – values over 100 stick to the top of the graph. The labels on the vertical axis are whatever you specify (but there’s no automatic relationship). I kludged together a defensible approach (values, which ranged from 5000-15000, divided by 160 to get everything below three digits and into a more or less correct relationship with the vertical axis) but a tidier approach would have been to make all values a percentage of the highest value in the data set, in this case about 15,000, and use that value to label the top of the y axis.
Here’s a map which has over 1,000 customized pie graphs.