That's a nice graph. It's good to see distance of closest defender considered.
Naively it looks like you should always take close in 2-pointers, because they have more expected points per shot than 3-pointers -- even comparing the purple graph for twos vs. the red graph for threes. But of course a lot more is going on:
You're less likely to have a close defender at the 3-point line because there's so much more territory for the defense to cover than around the basket. So 3-pointers with a distant defender are more likely to be available. (Midrange too, of course.)
There's a shot clock, so you can't work the ball around forever looking for the optimal shot.
A close defender facing you is not the same as a close defender you've gotten by, and a defender shorter than you who you can easily shoot over isn't the same as a taller one who will block your shot. So even this graph doesn't fully factor out what I called the selection bias of considering only shots actually taken.
Etc.
Naively it looks like you should always take close in 2-pointers, because they have more expected points per shot than 3-pointers -- even comparing the purple graph for twos vs. the red graph for threes. But of course a lot more is going on:
You're less likely to have a close defender at the 3-point line because there's so much more territory for the defense to cover than around the basket. So 3-pointers with a distant defender are more likely to be available. (Midrange too, of course.)
There's a shot clock, so you can't work the ball around forever looking for the optimal shot.
A close defender facing you is not the same as a close defender you've gotten by, and a defender shorter than you who you can easily shoot over isn't the same as a taller one who will block your shot. So even this graph doesn't fully factor out what I called the selection bias of considering only shots actually taken.
Etc.