@fusaroli @rlmcelreath (2/) Gamma from {shape,scale} to {log-mean, log-shape} is often good. In https://t.co/La6Y2SI4kF we suggest scaling; eliminating correlation; making contours elliptical. I would love to see more examples, especially about elim diverg
5,366 followers
5,366 followers
@geospacedman @DaniRabaiotti @pitakakariki Fair enough. I still have a hunch that R is particularly bad in this respect. Interesting but hard to try to quantify this reasonably rigorously (@gvwilson ?) Thought about this when writing this paper https://t.c
307 followers
Words of wisdom from past Trevor which I still agree with cc:@nceas Final product which was my privilege to be part of: https://t.co/mtqrR7vMXw https://t.co/X2I86HDdFq
8,495 followers
Fréd Gosselin talks about hierarchical models and the issues we face when fitting them; see e.g. https://t.co/0tyOCNAv9U, or https://t.co/8kLznaQPXW by @AugerMethe @AEDerocher @IanJonsen et al. and https://t.co/wIWgpFWGr7 by @bolkerb @LabGardner @AdmbProje