ML Output Kenya
Fiona Spooner
July 23, 2019
library(here)
library(dplyr)
library(ggplot2)
library(stringr)
library(reshape2)
library(purrr)
library(readr)
library(scales)Effort
Calculating the daily effort at each site i.e. the number of cameras that were on each day at each site
mn<-read.csv(here::here("effort/naboisho_effort.csv"), stringsAsFactors = FALSE, na.strings=c("","NA"))
mn$Start_1<-as.Date(mn$Start_1, format = "%d-%m-%y")
mn$End_1<-as.Date(mn$End_1, format = "%d-%m-%y")
mn$Start_2<-as.Date(mn$Start_2, format = "%d-%m-%y")
mn$End_2<-as.Date(mn$End_2, format = "%d-%m-%y")
start<-min(na.omit(mn$Start_1))
end<-max(na.omit(mn$End_2))
all_days<-seq(start, end, by = "day")
all_zeros<-rep(0, length(all_days))
effort_func<-function(loc){
site<-mn[mn$Location_ID == loc,]
effort<-all_zeros
if(!is.na(site$Start_1) & !is.na(site$End_1)){
loc_start1<-site$Start_1
loc_end1<-site$End_1
effort[which(all_days == loc_start1):which(all_days == loc_end1)]<-1
}
if(!is.na(site$Start_2) & !is.na(site$End_2)){
loc_start2<-site$Start_2
loc_end2<-site$End_2
effort[which(all_days == loc_start2):which(all_days == loc_end2)]<-1
}
print(loc)
return(effort)
}
effort_out<-lapply(mn$Location_ID,effort_func)
effort_mat<-do.call("rbind", effort_out)
effort_sum<-colSums(effort_mat)
plot(effort_sum, type = "l")
effort_df<-data.frame(mn$Location_ID,effort_mat)
colnames(effort_df)<-c("Location_ID", as.character(all_days))
#write.csv(effort_df, here::here("effort/naboisho_effort_matrix.csv"), row.names = FALSE)Combining the effort matrices
effort_files<-list.files(here::here("/effort"),pattern = "*_effort_matrix.csv")
eff<-read.csv(effort_files[4])
eff_sum<-melt(colSums(eff[,2:ncol(eff)]))
eff_sum$date<-as.Date(row.names(eff_sum), format = "X%Y.%m.%d")
eff_sum$site<-"OMC"
#write.csv(eff_sum, here::here("/effort/omc_daily_effort.csv"), row.names = FALSE)files <- list.files(here::here("effort"),pattern = "*daily_effort.csv", full.names = TRUE)
ef <- files %>%
map(read_csv) %>% # read in all the files individually, using
reduce(rbind) # reduce with rbind into one dataframe
colnames(ef)[1]<-"effort"
#write.csv(ef,here::here("effort/all_sites_effort.csv"), row.names = FALSE)Animal vs No Animal
Producing figures to show the total number of images at each site categorised as either Animal or no Animal
lf<-list.files(here::here("/results_csvs/phase1"), pattern = "*_results.csv", full.names = TRUE)
filer<-function(file){
file_in<-read.csv(file, header = FALSE)
file_in$filename<-basename(file)
return(file_in)
}
files_out<-lapply(lf, filer)
allp1<-do.call("rbind", files_out)
write.csv(allp1, here::here("results_csvs/phase1/all_sites_seasons.csv"), row.names = FALSE)allp1<-read.csv(here::here("results_csvs/phase1/all_sites_seasons.csv"), stringsAsFactors = FALSE)
colnames(allp1)<-c("row_no", "IMG_ID","unsure", "Animal", "NoAnimal", "ConfAnim", "ConfNoAnim", "SiteSeason")Formatting the ML output
allp1$Site<-allp1$SiteSeason%>%
gsub("_november_results.csv", "", .)%>%
gsub("_october_results.csv", "", .)%>%
gsub("_november_repaired_results.csv", "", .)%>%
gsub("_october_repaired_results.csv", "", .)
allp1$Season<-allp1$SiteSeason%>%
gsub("_results.csv", "", .)%>%
gsub("_repaired", "", .)%>%
gsub("mara_north_", "", .)%>%
gsub("mara_triangle_", "", .)%>%
gsub("naboisho_", "", .)%>%
gsub("omc_", "", .)
#getting rid of the summary rows at the end of each data sheet
alldf<-allp1[-which(allp1$Animal != "[0" & allp1$Animal != "[1"),]
alldf$Animal[alldf$Animal == "[0"]<-0
alldf$Animal[alldf$Animal == "[1"]<-1
alldf$ConfAnim<-gsub("\\[","", alldf$ConfAnim)
alldf$IMG_ID_nospl<-alldf$IMG_ID%>%
gsub("_0.JPG","", .)%>%
gsub("_1.JPG","", .)Merging the split images together by summing the animal column. Anything with a 1 or 2 will be recoded as 1, and 0 will remain 0.
ggplot(df_all, aes(SumAnimalF, fill = SumAnimalF))+
geom_bar()+
facet_wrap(.~Site_f, labeller = labeller(Site_f = site_labs))+
labs(x = "", y = "Number of Images")+
scale_x_discrete(labels=c("0" = "No Animal", "1" = "Animal"))+
scale_y_continuous(labels = comma)+
scale_fill_manual( values = c("0" = "black", "1" = "grey"))+
theme_bw()+
theme(axis.text.y = element_text(lineheight = 0.5 , size = 18),
axis.text.x = element_text(size = 14),
axis.title.x = element_text(size = 20),
axis.title.y = element_text(size = 20),
legend.position = "none",
strip.text.x = element_text(size = 14))
sitesum<-df_all %>%
group_by(Site_f,SumAnimalF, SiteCode)%>%
#filter(MeanConf >= 0.9)%>%
summarise(Freq = n())
sitesum$SumAnimalF<-as.factor(sitesum$SumAnimalF)
sitesum$SumAnimalF<-factor(sitesum$SumAnimalF, levels(sitesum$SumAnimalF)[c(2,1)])
ggplot(sitesum, aes(x =Site_f ,y = Freq ,alpha = SumAnimalF, fill = Site_f))+
geom_bar(position="dodge", stat="identity")+
labs(x = "", y = "Number of Images", fill ="", alpha = "")+
scale_alpha_discrete(range = c(1, 0.35), labels = c("Animal", "No Animal"))+
scale_x_discrete(labels=c("mara_north" = "Mara North", "mara_triangle" = "Mara Triangle", "naboisho" = "Naboisho", "omc" = "OMC"))+
guides(fill = FALSE)+
scale_y_continuous(labels = comma)+
theme_bw()+
#scale_fill_manual(values = c("0" = "black", "1" = "grey"), labels = c("No Animal", "Animal"))+
theme(axis.text.y = element_text(lineheight = 0.5 , size = 18),
axis.text.x = element_text(size = 18),
axis.title.x = element_text(size = 20),
axis.title.y = element_text(size = 20),
legend.text = element_text(size = 18))
#write.csv(sitesum, "Kenya_Animal_No_Animal_Summary.csv", row.names = FALSE)Image counts with effort accounted for
ef<-read.csv(here::here("effort/all_sites_effort.csv"), stringsAsFactors = FALSE)
ef_sum <-ef %>%
group_by(site)%>%
summarise(sum_ef = sum(effort))
ef_tot<-data.frame("Total",sum(ef$effort) )
colnames(ef_tot)<-colnames(ef_sum)
ef_all<-rbind(ef_tot, ef_sum)ggplot(site_ef, aes(x =Site_f ,y = FreqEff ,alpha = SumAnimalF, fill = Site_f))+
geom_bar(position="dodge", stat="identity")+
labs(x = "", y = "Number of Images per Camera Trap Day", fill ="", alpha = "")+
scale_alpha_discrete(range = c(1, 0.35), labels = c("Animal", "No Animal"))+
scale_x_discrete(labels=c("mara_north" = "Mara North", "mara_triangle" = "Mara Triangle", "naboisho" = "Naboisho", "omc" = "OMC"))+
scale_y_continuous(labels = comma)+
guides(fill = FALSE)+
theme_bw()+
#scale_fill_manual(values = c("0" = "black", "1" = "grey"), labels = c("No Animal", "Animal"))+
theme(axis.text.y = element_text(lineheight = 0.5 , size = 18),
axis.text.x = element_text(size = 18),
axis.title.x = element_text(size = 20),
axis.title.y = element_text(size = 20),
legend.text = element_text(size = 18))
Excluding images below 90% Confidence Threshold
90% Confidence Threshold not accounting for effort
90% Confidence Threshold accounting for effort
ggplot(site_ef, aes(x =Site_f ,y = FreqEff ,alpha = SumAnimalF, fill = Site_f))+
geom_bar(position="dodge", stat="identity")+
labs(x = "", y = "Number of Images\n per Camera Trap Day", fill ="", alpha = "")+
scale_alpha_discrete(range = c(1, 0.35), labels = c("Animal", "No Animal"))+
scale_x_discrete(labels=c("mara_north" = "Mara North", "mara_triangle" = "Mara Triangle", "naboisho" = "Naboisho", "omc" = "OMC"))+
scale_y_continuous(labels = comma)+
guides(fill = FALSE)+
theme_bw()+
#scale_fill_manual(values = c("0" = "black", "1" = "grey"), labels = c("No Animal", "Animal"))+
theme(axis.text.y = element_text(lineheight = 0.5 , size = 18),
axis.text.x = element_text(size = 18),
axis.title.x = element_text(size = 20),
axis.title.y = element_text(size = 20),
legend.text = element_text(size = 18))
Species Counts
allp2<-read.csv(here::here("results_csvs/phase2/all_sites_seasons.csv"), stringsAsFactors = FALSE)
allp2<-dplyr::select(allp2, "V2", "V11", "V16", "filename")
colnames(allp2)<-c("IMG_ID", "Species_Code", "Confidence", "Filename")species_id<-read.csv("species_codes.csv", header = FALSE)
colnames(species_id)<-c("CommonName", "Code")allp2$Species_Code<-gsub("\\[", "", allp2$Species_Code)
allp2$Confidence<-gsub("\\[", "", allp2$Confidence)
allp2$Site<-allp2$Filename%>%
gsub("_november_phase2ready_phase2_results.csv", "", .)%>%
gsub("_october_phase2ready_phase2_results.csv", "", .)%>%
gsub("_november_repaired_phase2ready_phase2_results.csv", "", .)%>%
gsub("_october_repaired_phase2ready_phase2_results.csv", "", .)
allp2$Season<-allp2$Filename%>%
gsub("_phase2ready_phase2_results.csv", "", .)%>%
gsub("_repaired", "", .)%>%
gsub("mara_north_", "", .)%>%
gsub("mara_triangle_", "", .)%>%
gsub("naboisho_", "", .)%>%
gsub("omc_", "", .)
allp2$IMG_ID<-as.character(allp2$IMG_ID)
cam_split<-strsplit(allp2$IMG_ID, "/")
get_first<-function(x){
image_out<-x[[1]]
return(image_out)
}
cam_out<-lapply(cam_split,get_first)
allp2$CamUnique<-unlist(cam_out)Joining the ML output with the exif data to join the species data with a date
dfs<-allsp%>%
dplyr::select(ID_unique, Species_Code, Site, CamUnique, Season, CommonName, MeanConf)%>%
distinct()%>%
group_by(CommonName,Site)%>%
#filter(MeanConf >= 0.9)%>%
mutate(FreqCommonName = n())%>%
arrange(-FreqCommonName)
dfs$site_code <- dfs$Site
dfs$site_code[dfs$site_code == "naboisho"]<-"NB"
dfs$site_code[dfs$site_code == "mara_north"]<-"MN"
dfs$site_code[dfs$site_code == "mara_triangle"]<-"MT"
dfs$site_code[dfs$site_code == "omc"]<-"OMC"gg_color_hue <- function(n) {
hues = seq(15, 375, length = n + 1)
hcl(h = hues, l = 65, c = 100)[1:n]
}
cols = gg_color_hue(5)
site_code = c("Total", "MT", "OMC", "NB", "MN")
site_cols<-data.frame(cols, site_code)
df_col<-merge(df_all, site_cols, by = "site_code")
df_col$CommonName<-str_trim(df_col$CommonName)Total
Total, Confidence >= 90%
Mara North
Mara North Confidence >= 90%
Mara Triangle
Mara Triangle Confidence >= 90%
Naboisho
Naboisho Confidence >= 90%
OMC
OMC Confidence >= 90%
Site based species counts
Creating time series of effort for each conservancy
Images/Effort over time
ggplot(ef_all, aes(x = date, y = caps_per_ef, group = site_f, colour = site_f))+
geom_line(size =1.5)+
#geom_point()+
facet_grid(.~site_f, labeller = labeller(site_f = site_labs))+
labs(x = "Date", y = "Number of Images\nper Camera Trap Day", colour = "Site")+
scale_fill_discrete(labels = c("Total", "Mara Triangle", "OMC", "Naboisho", "Mara North"))+
theme_bw()+
theme(axis.text.y = element_text(lineheight = 0.5 , size = 18),
axis.text.x = element_text(size = 14),
axis.title.x = element_text(size = 20),
axis.title.y = element_text(size = 20),
legend.text = element_text(size = 18),
legend.title = element_text(size = 20),
strip.text.x = element_text(size = 18))
ggplot(ef_all, aes(x = date, y = caps_per_ef, group = site_f, fill = site_f))+
geom_bar(width=1,stat = "identity")+
facet_grid(.~site_f, labeller = labeller(site_f = site_labs))+
labs(x = "Date", y = "Number of Images\nper Camera Trap Day", fill = "Site")+
scale_fill_discrete(labels = c("Total", "Mara Triangle", "OMC", "Naboisho", "Mara North"))+
theme_bw()+
theme(axis.text.y = element_text(lineheight = 0.5 , size = 18),
axis.text.x = element_text(size = 14),
axis.title.x = element_text(size = 20),
axis.title.y = element_text(size = 20),
legend.text = element_text(size = 18),
legend.title = element_text(size = 20),
strip.text.x = element_text(size = 18))