Hi there,
So I see this and from some testing and The Github info on it. For the first Question ,YES/NO you could calibrate the space by doing a couple things. I have found the best results depend on the environment space. As you see when you open the buffer and walk to it the returns from the radar doppler effect on the target (you) the data will increase pretty logarithmically btw,
Now if you characterize the room and Know some things you can parse through the data and get some basic stuff right away.
EMPTY room (no People, Pets or fans, AC OFF) take some samples and see if it settles to a consistent baseline number. Call that your ZERO.
Then add the other stuff and get some samples for different scenarios Collect the numbers.
As @liaifat85 indicated, with good data points you can tell if it’s occupied and by ONE or MORE* but I don’t think a count is possible unless you could isolate more info.
Read the Git for more specific data.
HTH
GL 
PJ 
ps. even the particles in the space will effect the raw data when the buffer is open so sample more often and take less data in some cases. 