search

weka511 / learn

Experiments in machine learning
GNU General Public License v3.0
0 stars 0 forks source link

Handle 4 GCP peaks #4

Closed weka511 closed 4 years ago

weka511 commented 4 years ago

Some wells such as PAP15100449r H12 have 4 peaks. Perhaps we could try 3 or 4 peaks, then use the one with the maximum likelihood.

weka511 commented 4 years ago

Standard deviation may be a better clue: it looks as if an incorrect K gives rise to at least one high value of sigma

weka511 commented 4 years ago

diffs.txt

weka511 commented 4 years ago

I've eyeballed the plots, and there are a few ones where K has been estimated wrongly: PAP90200082-H12 (is this the only one?)