I have some of these units as motion detectors in a home sentry system. (I bought the raw radar transceiver and designed the amplifier/square wave generator, but the Parallax unit should give nearly the same results.) They are very sensitive, and should easily detect a basketball moving toward or away from them. Remember that a Doppler unit detects relative motion. Therefore, if the unit itself is not stationary, it's going to generate an output.
If you have attached this to the bottom of the actual net, won't it move when the wind blows? Your best position for this, I think, given the many angles at which the ball can approach, would be top center of the backboard, angled down by some amount. You aint going to win with this unless its mount is maximally stationary.
And as for detection, remember that the faster something is approaching (or moving away), the more square waves per second it generates. So a ball is moving pretty fast, like maybe 5 mph or more (just a guess), which should generate frequencies say in the 50 - 150 Hz. range. I set mine up so the square waves trigger an interrupt, which just counts incoming square waves. If you then set up a timer to start the count over, say every half second, you can look for enough square waves to be sure that your reading isn't ambient noise. Just an approximation, but each mph of closing/opening motion (i.e., directly at/away from the detector), will deliver about 10Hz.
I set mine up to ignore less than 0.5 mph (related to people walking), and it has yet to ever false alarm.