Changes

Jump to navigation Jump to search
#A (set of) comparison(s) between the max R2 method and the elbow methods
#A new heatmap or two, based on a different location.
 
I also checked the implementation of the % layer index again, and fixed a mistake in it.
 
In Heurcalc, I define fracunclusted as:
CASE WHEN finallayer >1 THEN ((layer::real-1)/(finallayer::real-1)) ELSE NULL END AS fracunclustered,
 
I then previously select the layer below or equal to % layer index 0.374833 in Heurflh:
CASE WHEN floor((finallayer-1)*0.374833)=0 THEN 1::int ELSE floor((finallayer-1)*0.374833) END AS heurflhlayer
 
This was incorrect. With some rearrangement, we can see:
<math>index \le 0.374833 \implies \frac{layer-1}{final-1} \le 0.374833 \implies layer \le 0.374833(final-1) + 1 </math>
 
I therefore fixed the query to:
floor(((finallayer-1)*0.374833)+1) AS heurflhlayer
====New Do File====

Navigation menu