attrition (i should mention it is in my vocabulary the technical type) happened again today on A5, and the two shrinking EBs flagged orange have been excluded computationally and soon be thrown away before the transition.
this legion is tighter and tighter as the anabasis continues.
- 50 is the minimum viable threshold for cv. i’m at 82. 82 to 50 would only happen from catastrophe not attrition or biostochasticism.
- the strategy for post-transfer is worthy to plan around. the rbqc will be affected (especially when screening for edge effects since layout completely changes); rbqc gui (requires DRASTIC modifications for better legibility); covariates screening will require two versions due to spatial variation at two different temporal ranges (before and after d21); dataset population (perhaps just reverse-map to old to preserve continuity); anything upstream and during segmentation. in short, total engineering overhaul.
to counter lack of data is to make engineering less crude and more precise.
we will see how much we are going to lose.
b says edge effects only affect seeding stage, which is early. but covariate is still a good and necessary computational insurance since she doesn’t implement any physical insurance (ie buffers).