Hierarchical generalized additive models in ecology: an introduction with mgcv

View article
1351 days ago
@JohsEnevoldsen @mattansb
reply repost on X like
1403 days ago
@stephenjwild @IsabellaGhement Fig 4 of Pedersen et al. “Hierarchical Generalized Additive Models in Ecology: An Introduction with Mgcv.” PeerJ 2019 is really useful here: https://t.co/nCZniae3Ab @ericJpedersen @noamross @ucfagls @millerdl https://t.co/ZDbVDpGEh8
reply repost on X like
1406 days ago
I love papers like this - written in such a way to be directly useful for people to start employing these tools in their work. https://t.co/VW1RElLu4i @noamross et al.
reply repost on X like
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
1443 days ago
RT @StaffanBetner: @SolomonKurz Suggestion: dv ~ group + s(time, by = group) + s(time, participant, bs = "fs") group and participant need…
reply repost on X like
1444 days ago
@bmwiernik @SolomonKurz Agreed, here is a great overview from @ucfagls and colleagues https://t.co/V56yz7UA8W
reply repost on X like
@SolomonKurz Suggestion: dv ~ group + s(time, by = group) + s(time, participant, bs = "fs") group and participant need to be coded as factors Suggested reading: https://t.co/AJRoemPMmp (by @ericJpedersen @millerdl @ucfagls @noamross)
reply repost on X like
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
1480 days ago
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
@IsabellaGhement isnt that similar to model "GS" in this article? https://t.co/iEoKRMLYvQ global smooth, with group level smooths each with the same penalty
reply repost on X like
@GuidoBiele We discuss this in the HGAM paper https://t.co/SZEBia7QYa but the problem is more your choice of the model, yours seems over specified or unidentifiable
reply repost on X like
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
@ucfagls @stephenjwild @PhDemetri @bolkerb Gavin, for HGAM model comparison, the paragraph in your HGAM article https://t.co/NtuencHy5R does talk about model comparison. Does the paragraph refer solely to prediction settings for that comparison (with method = "REML" to make comparisons meaningful)? https://t.co/HX5YXGs5MP
reply repost on X like
1544 days ago
RT @ucfagls: If you want to fit nonlinear relationships to data where those relationships can vary by groups (eg species) then we have a pa…
reply repost on X like
@bmwiernik You can get (and plot) the basis functions mgcv uses using basis() and the penalty matrix via penalty() and their draw() methods from my {gratia}
reply repost on X like
Since not only detection can change but also potentially density (if we have surveys of birds on the water and birds in flight as our two "platforms"), we adapt the factor-smooth approach (https://t.co/asA2WQhXhG) to account for this https://t.co/lr73CHvwNs
reply repost on X like
@Toffeemen68 @ali__johnston set.seed(11) sample(c('Miller','Ross','Simpson')) here https://t.co/rRhyoFZUDa - @ericJpedersen @ucfagls @noamross @millerdl
reply repost on X like
@andrewheiss @tjmahr Ooo I'll check this out. I came across this on a Stan message board and they draw parallels between the two, but make smooths sound like a more flexible/less group focused approach. Or at least you control "group" boundaries more directly? https://t.co/WROcriUBtQ
reply repost on X like
RT @fbuderman: @SophieLGilbert @kaitlyngaynor @cole_burton @JasonTFisherLab @RLongEco @LabGardner @taaltree @Fabi_Iannarilli @songofdodo Ba…
reply repost on X like
RT @fbuderman: @SophieLGilbert @kaitlyngaynor @cole_burton @JasonTFisherLab @RLongEco @LabGardner @taaltree @Fabi_Iannarilli @songofdodo Ba…
reply repost on X like