data & packages
df = readRDS("ecd_obs_agathe.RDS")
df <- df[!is.na(df$obs_ts),]
df[is.na(df)] <- 0
library('mgcv')
## Loading required package: nlme
## This is mgcv 1.8-34. For overview type 'help("mgcv-package")'.
library('mgcViz')
## Loading required package: qgam
## Loading required package: ggplot2
## Loading required package: rgl
## This build of rgl does not include OpenGL functions. Use
## rglwidget() to display results, e.g. via options(rgl.printRglwidget = TRUE).
## Registered S3 method overwritten by 'GGally':
## method from
## +.gg ggplot2
## Registered S3 method overwritten by 'mgcViz':
## method from
## +.gg GGally
##
## Attaching package: 'mgcViz'
## The following objects are masked from 'package:stats':
##
## qqline, qqnorm, qqplot
library('ggpubr')
## Registered S3 methods overwritten by 'car':
## method from
## influence.merMod lme4
## cooks.distance.influence.merMod lme4
## dfbeta.influence.merMod lme4
## dfbetas.influence.merMod lme4
Species to consider + Stratification+ Legende+ Distribution
espece<-c("num_Other_bony_fishes","num_Silky_shark","num_Common_dolphinfish")
df$num_Other_bony_fishes=df$`num_Other bony fishes`
df$num_Common_dolphinfish=df$`num_Common dolphinfish`
df$num_Silky_shark=df$`num_Silky shark`
Stratification & legende & distribution
df$group = paste0("O",df$ocean,"G",df$code_assoc_groupe)
# Création d'une variable groupe
# O1G1, O1G2, O2G1, O2G2
# O1: Atlantique 02: Indien
# G1: Floating object G2: Free swiming school
## Par océan et banc
df1<-df[df$group =="O1G1",]
# df1 contient tous les calais qui sont sont O1G1
df2<-df[df$group =="O1G2",]
# df2 contient tous les calais qui sont O1G2
df3<-df[df$group =="O2G1",]
# df3 contient tous les calais qui sont O2G1
df4<-df[df$group =="O2G2",]
# df4 contient tous les calais qui sont O2G2
# légende -----------------------------------------------
strat<-c("O1G1","O1G2","O2G1","O2G2")
data_list<-list(O1G1=df1,O1G2=df2,O2G1=df3,O2G2=df4)
# Distribution----------------------------------------
famille<-"gaussian"
Results GAM
res_gamcapture<-list()
for (i in 1:length(espece)){
for (j in 1:length(data_list)){
data <- data_list[[j]]
form <- formula(paste(espece[i],"s(capture_total_target_tunas_corrigee,k=10)", sep="~"))
gam <- gam(form, data=data)
gr=strat[j]
sp=gsub("^num_","",espece[i])
fn=paste(gr,sp,sep=".")
res_gamcapture[[paste(gr,sp,sep=".")]] = list(sp=sp,gr=gr,fn=fn,gam=gam)
}
}
GAM for Atlantic ocean , Floating object
gam_O1G1<-list(res_gamcapture$O1G1.Other_bony_fishes,res_gamcapture$O1G1.Silky_shark,res_gamcapture$O1G1.Common_dolphinfish)
for (v in gam_O1G1){
#Diagnostic
par(mfrow=c(2,2))
gam.check(v$gam)
# plot gam
o <- getViz(v$gam)
plot <- plot(o) + l_fitLine(colour = "red") + l_rug(mapping = aes(x=x, y=y), alpha = 0.8) +
l_ciLine(mul = 5, colour = "blue", linetype = 2) +
l_points(shape = 19, size = 1, alpha = 0.1) + theme_classic()+
ggtitle(v$fn)+
ylab(v$esp)+
xlab("capture total target tunas corrigee (weight in tonnes)")
print(plot)
}
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 7 iterations.
## The RMS GCV score gradient at convergence was 1.736867e-07 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9.00 6.22 0.92 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 4 iterations.
## The RMS GCV score gradient at convergence was 1.33085e-07 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9.00 3.85 1 0.38
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 4 iterations.
## The RMS GCV score gradient at convergence was 6.352695e-06 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9.00 2.17 0.95 0.005 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
GAM for Atlantic ocean, free school
gam_O1G2<-list(res_gamcapture$O1G2.Other_bony_fishes,res_gamcapture$O1G2.Silky_shark,res_gamcapture$O1G2.Common_dolphinfish)
for (v in gam_O1G2){
#Diagnostic
par(mfrow=c(2,2))
gam.check(v$gam)
# plot gam
o <- getViz(v$gam)
plot <- plot(o) + l_fitLine(colour = "red") + l_rug(mapping = aes(x=x, y=y), alpha = 0.8) +
l_ciLine(mul = 5, colour = "blue", linetype = 2) +
l_points(shape = 19, size = 1, alpha = 0.1) + theme_classic()+
ggtitle(v$fn)+
ylab(v$esp)+
xlab("capture total target tunas corrigee (weight in tonnes)")
print(plot)
}
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 8 iterations.
## The RMS GCV score gradient at convergence was 7.132046e-08 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9 1 1.01 0.8
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 5 iterations.
## The RMS GCV score gradient at convergence was 2.884136e-06 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9.00 2.48 0.96 0.005 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 5 iterations.
## The RMS GCV score gradient at convergence was 1.246303e-07 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9 1 1 0.43
GAM for Indian ocean, floating object
gam_O2G1<-list(res_gamcapture$O2G1.Other_bony_fishes,res_gamcapture$O2G1.Silky_shark,res_gamcapture$O2G1.Common_dolphinfish)
for (v in gam_O2G1){
#Diagnostic
par(mfrow=c(2,2))
gam.check(v$gam)
# plot gam
o <- getViz(v$gam)
plot <- plot(o) + l_fitLine(colour = "red") + l_rug(mapping = aes(x=x, y=y), alpha = 0.8) +
l_ciLine(mul = 5, colour = "blue", linetype = 2) +
l_points(shape = 19, size = 1, alpha = 0.1) + theme_classic()+
ggtitle(v$fn)
#ylab(v$esp)+
#xlab("capture total target tunas corrigee (weight in tonnes)")
print(plot)
}
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 7 iterations.
## The RMS GCV score gradient at convergence was 2.503706e-05 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9.00 5.38 0.93 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 5 iterations.
## The RMS GCV score gradient at convergence was 5.711209e-06 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9.00 5.09 0.93 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 4 iterations.
## The RMS GCV score gradient at convergence was 3.214313e-08 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9.00 4.41 0.96 0.005 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
GAM for Indian ocean, free school
gam_O2G2<-list(res_gamcapture$O2G2.Other_bony_fishes,res_gamcapture$O2G2.Silky_shark,res_gamcapture$O2G2.Common_dolphinfish)
for (v in gam_O2G2){
#Diagnostic
par(mfrow=c(2,2))
gam.check(v$gam)
# plot gam
o <- getViz(v$gam)
plot <- plot(o) + l_fitLine(colour = "red") + l_rug(mapping = aes(x=x, y=y), alpha = 0.8) +
l_ciLine(mul = 5, colour = "blue", linetype = 2) +
l_points(shape = 19, size = 1, alpha = 0.1) + theme_classic()+
ggtitle(v$fn)+
ylab(v$esp)+
xlab("capture total target tunas corrigee (weight in tonnes)")
print(plot)
}
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 4 iterations.
## The RMS GCV score gradient at convergence was 1.207353e-07 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9.00 3.93 1.01 0.44
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 8 iterations.
## The RMS GCV score gradient at convergence was 6.280597e-08 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9 1 0.99 0.18
##
## Method: GCV Optimizer: magic
## Smoothing parameter selection converged after 4 iterations.
## The RMS GCV score gradient at convergence was 1.286937e-06 .
## The Hessian was positive definite.
## Model rank = 10 / 10
##
## Basis dimension (k) checking results. Low p-value (k-index<1) may
## indicate that k is too low, especially if edf is close to k'.
##
## k' edf k-index p-value
## s(capture_total_target_tunas_corrigee) 9 1 0.96 0.07 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
