Mercurial > repos > ecology > vigiechiro_idcorrect_2ndlayer
view IdCorrect_2ndLayer.R @ 0:6681b6ba1d7e draft
planemo upload for repository https://github.com/galaxyecology/tools-ecology/tools/vigiechiro commit d2de8e10c11bfa3b04729e59bba58e08d23b56aa
author | ecology |
---|---|
date | Wed, 13 Mar 2019 11:18:36 -0400 |
parents | |
children |
line wrap: on
line source
#!/usr/bin/env Rscript suppressMessages(library(data.table)) suppressMessages(library(randomForest)) args <- commandArgs(trailingOnly = TRUE) set.seed(1) #To test reproductibility filename=args[3] if (exists("ClassifEspC2b")==F){load(args[2])} DataPar=fread(args[1],na.strings="") #id to be corrected DataPar$participation=substr(filename,nchar(filename)-40,nchar(filename)-17) test1=duplicated(cbind(DataPar$'nom du fichier',DataPar$tadarida_taxon)) test2=(DataPar$tadarida_taxon=="empty") DataPar=subset(DataPar,(!test1)|(test2)) DataPar$tadarida_probabilite[DataPar$tadarida_probabilite==""]="0" DataPar$tadarida_probabilite=as.numeric(DataPar$tadarida_probabilite) #table counting number of contacts per species nbcT=as.matrix(table(DataPar$participation,DataPar$tadarida_taxon)) DataPar$tadarida_probabilite=as.numeric(DataPar$tadarida_probabilite) #generating input variables for second layer classification Q25=vector() Q50=vector() Q75=vector() Q90=vector() Q95=vector() Q98=vector() Q100=vector() compt=0 PropSp=nbcT[0,] VoteO=DataPar[0,] for (j in 1:nlevels(as.factor(DataPar$tadarida_taxon))) { Datasub2=subset(DataPar,DataPar$tadarida_taxon==levels(as.factor(DataPar$tadarida_taxon))[j]) Q25=c(Q25,rep(quantile(Datasub2$tadarida_probabilite,0.25),nrow(Datasub2))) Q50=c(Q50,rep(quantile(Datasub2$tadarida_probabilite,0.50),nrow(Datasub2))) Q75=c(Q75,rep(quantile(Datasub2$tadarida_probabilite,0.75),nrow(Datasub2))) Q90=c(Q90,rep(quantile(Datasub2$tadarida_probabilite,0.90),nrow(Datasub2))) Q95=c(Q95,rep(quantile(Datasub2$tadarida_probabilite,0.95),nrow(Datasub2))) Q98=c(Q98,rep(quantile(Datasub2$tadarida_probabilite,0.98),nrow(Datasub2))) Q100=c(Q100,rep(max(Datasub2$tadarida_probabilite),nrow(Datasub2))) Ncont1=nrow(Datasub2) VoteO=rbind(VoteO,Datasub2) PropSp0=nbcT/Ncont1 PropSp=rbind(PropSp,PropSp0[rep(seq_len(nrow(PropSp0)),nrow(Datasub2)),]) compt=compt+nrow(Datasub2) #print(paste(compt,levels(as.factor(DataPar$tadarida_taxon))[j])) } VoteC2=cbind(VoteO,PropSp,Q25,Q50,Q75,Q90,Q95,Q98,Q100) #editing column titles to identify var of type "proportion d'abondances" for (i in 15:(ncol(VoteC2)-7)) { colnames(VoteC2)[i]=paste0(names(VoteC2)[i],"_prop") } #Add missing species EspForm=subset(row.names(ClassifEspC2b$importance) ,substr(row.names(ClassifEspC2b$importance) ,nchar(row.names(ClassifEspC2b$importance))-4 ,nchar(row.names(ClassifEspC2b$importance))) =="_prop") test=match(EspForm,colnames(VoteC2)) EspM=subset(EspForm,is.na(test)) Zeros=matrix(nrow=nrow(VoteC2),ncol=length(EspM)) Zeros[is.na(Zeros)]=0 colnames(Zeros)=EspM VoteC2=cbind(VoteC2,Zeros) ListDV=levels(as.factor(DataPar$'nom du fichier')) #calcule les probabilités max par espèce et par fichier #(utile pour corriger les erreurs dues à la coexistence de taxons dans le même fichier #ex: cris sociaux de Pipistrelles identifiées comme autre chose (Noctule, oreillard...)) #comptue max proba per species and files #(useful to correct errors that came from multiple taxons in the same file #eg ; Pipistrelles socials shouting identified as something else (Noctule, oreillard..)) MaxI=tapply(DataPar$tadarida_probabilite ,INDEX=list(c(DataPar$'nom du fichier'),c(DataPar$tadarida_taxon)) ,FUN=max) MaxI2=as.data.frame(cbind(row.names(MaxI),MaxI)) for (i in 2:ncol(MaxI2)) { MaxI2[,i]=as.numeric(as.character(MaxI2[,i])) } MaxI2[is.na(MaxI2)]=0 #édition des titres de colonne pour identifier les variables de type "indices max" #editing col titles to identify "indices max" variables for (i in 2:(ncol(MaxI2))) { colnames(MaxI2)[i]=paste0(names(MaxI2)[i],"_maxI") } #add missing species EspForm=subset(row.names(ClassifEspC2b$importance) ,substr(row.names(ClassifEspC2b$importance) ,nchar(row.names(ClassifEspC2b$importance))-4 ,nchar(row.names(ClassifEspC2b$importance))) =="_maxI") test=match(EspForm,colnames(MaxI2)) EspM=subset(EspForm,is.na(test)) Zeros=matrix(nrow=nrow(MaxI2),ncol=length(EspM)) Zeros[is.na(Zeros)]=0 colnames(Zeros)=EspM MaxI2=cbind(MaxI2,Zeros) #indice de confiance à l'echelle de l'observation (groupe de cris identifié comme provenant d'une seule espèce par la première couche) #Confidence indice on obs scale (shoutings groups identified as comming from a single species from the first layer) if(exists("IdS3")){rm(IdS3)} for (i in 1:nlevels(as.factor(DataPar$tadarida_taxon))) { Idsub=subset(DataPar,DataPar$tadarida_taxon==levels(as.factor(DataPar$tadarida_taxon))[i]) IdS2=cbind('nom du fichier'=Idsub$'nom du fichier',tadarida_taxon=Idsub$tadarida_taxon,prob=Idsub$tadarida_probabilite) colnames(IdS2)[3]=paste(levels(as.factor(DataPar$tadarida_taxon))[i]) if(exists("IdS3")){IdS3=merge(IdS3,IdS2,all=T)}else{IdS3=IdS2} } for (i in 3:ncol(IdS3)) { IdS3[,i]=as.numeric(as.character(IdS3[,i])) } #édition des titres de colonne pour identifier les variables de type "indices de l'observation" #editing col titles to identify "indices de l'observation" variables for (i in 3:(ncol(IdS3))) { colnames(IdS3)[i]=paste0(names(IdS3)[i],"_ValI") } IdS3[is.na(IdS3)]=0 #add missing species EspForm=subset(row.names(ClassifEspC2b$importance) ,substr(row.names(ClassifEspC2b$importance) ,nchar(row.names(ClassifEspC2b$importance))-4 ,nchar(row.names(ClassifEspC2b$importance))) =="_ValI") test=match(EspForm,colnames(IdS3)) EspM=subset(EspForm,is.na(test)) Zeros=matrix(nrow=nrow(IdS3),ncol=length(EspM)) Zeros[is.na(Zeros)]=0 colnames(Zeros)=EspM IdS3=cbind(IdS3,Zeros) #on merge les prop d'espèces, les quantiles et les indices par fichiers et par observations #merge species probabilities, quantiles and indice per files and per obs VoteC3=merge(VoteC2,MaxI2,by.x="nom du fichier",by.y="V1") VoteC4=merge(VoteC3,IdS3,by=c("nom du fichier","tadarida_taxon")) VoteC4$temps_fin=as.numeric(as.character(VoteC4$temps_fin)) VoteC4$temps_debut=as.numeric(as.character(VoteC4$temps_debut)) VoteC4$frequence=as.numeric(as.character(VoteC4$frequence_mediane)) VoteC4$durseq=VoteC4$temps_fin-VoteC4$temps_debut ProbEsp_C2b=predict(ClassifEspC2b,VoteC4,type="prob",norm.votes=TRUE) ProbEsp_C2bs=predict(ClassifEspC2b,VoteC4,type="response",norm.votes=TRUE) colnum=match("participation",colnames(VoteC4)) DataCorrC2=cbind(VoteC4[,1:colnum],ProbEsp_C2b,ProbEsp_C2bs) DataCorrC2=DataCorrC2[order(DataCorrC2$tadarida_probabilite,decreasing=T),] DataCorrC2=DataCorrC2[order(DataCorrC2$'nom du fichier'),] DataCorrC2$ProbEsp_C2bs=as.character(DataCorrC2$ProbEsp_C2bs) DataCorrC2$ProbEsp_C2bs[is.na(DataCorrC2$ProbEsp_C2bs)]="empty" fout_name="output.tabular" write.table(DataCorrC2,file=fout_name,row.names=FALSE,sep="\t",quote=FALSE,na="NA")