Trimming a dataset to best represent range?

Discussion created by occidentalis on Sep 25, 2012

I have a question that is difficult for me to find information on, since I'm not exactly sure what I should be searching. Forgive me if I've posted it in the wrong place. I have a dataset that covers North America, with about 500 data points, many of which are clustered together because of the way sampling occurred. For further analysis, I need to select 80 or so of those points that best spatially represent the dataset, so points that are furthest away from each other but still cover North America as completely as possible. How would I go about doing this?

Many thanks - even if you don't have an answer for me, any input on search terms would be great. I can't be the first with this task in mind.