R/teste-computacional-carga-dados.R

In mcf-rocha/iplan: iPlan

library(logging)
basicConfig()
addHandler(writeToFile, file="~/iplan.log")
#for (n in seq(from = 1, to = 10, by =1)) {
#######################
# n<-45
# ciclosEntrega<-1
# duracaoCiclo<-10
#
#n<-32
ID_execucao<-1
n<-60
ciclosEntrega<-2
duracaoCiclo<-15
juro<-0.003 # A SELIC em Dez/17 foi 0,54% = 0,0054
juroBeneficioTangivel<-0.003
rm(list=setdiff(ls(), c("n","duracaoCiclo","ciclosEntrega","juro", "juroBeneficioTangivel")))
#######################
source("~/projetosR/iplan/R/iplan.R")
loginfo(paste("############ INICIANDO RODADA CONTENDO ",n," FUNCIONALIDADES ###############",sep = ""))
inicio.time <- Sys.time()

##################################################
epico.projeto<-NULL
prazo<-NULL
avaliacao<-NULL
projeto<-NULL
escala<-c("Insignificante","Pequena","Moderada","Alta","Extrema")

for (j in 1:3) {
  epico.projeto<-c(epico.projeto,"F")
  prazo<-c(prazo,paste0("C",j))
  avaliacao<-c(avaliacao,sample(escala,1,T))
  projeto<-c(projeto,"")
}

for (i in 1:n) {
  for (j in 1:3) {
    epico.projeto<-c(epico.projeto,paste0("f",i))
    prazo<-c(prazo,paste0("C",j))
    avaliacao<-c(avaliacao,sample(escala,1,T))
    projeto<-c(projeto,"F")
  }
}

iplan_beneficios_intangiveis_avaliacao <- data.frame(epico.projeto,prazo,avaliacao,projeto)

write.csv(iplan_beneficios_intangiveis_avaliacao,
          file = "~/Doutorado/iplan-teste-computacional/iplan-beneficios-intangiveis-avaliacao.csv",
          quote = FALSE,
          row.names=FALSE)
###################################################
Origem<-NULL
Destino<-NULL
#-------------------------------
# anterior<-"begin"
# for (j in seq(1,5,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",j))
#   anterior<-paste0("f",j)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (j in seq(6,10,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",j))
#   anterior<-paste0("f",j)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (j in seq(11,15,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",j))
#   anterior<-paste0("f",j)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (j in seq(16,20,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",j))
#   anterior<-paste0("f",j)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (j in seq(21,25,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",j))
#   anterior<-paste0("f",j)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (j in seq(26,30,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",j))
#   anterior<-paste0("f",j)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (k in seq(31,35,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",k))
#   anterior<-paste0("f",k)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (k in seq(36,40,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",k))
#   anterior<-paste0("f",k)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (k in seq(41,45,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",k))
#   anterior<-paste0("f",k)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (k in seq(46,50,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",k))
#   anterior<-paste0("f",k)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (k in seq(51,55,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",k))
#   anterior<-paste0("f",k)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")
# 
# anterior<-"begin"
# for (k in seq(56,60,1)) {
#   Origem<-c(Origem,anterior)
#   Destino<-c(Destino,paste0("f",k))
#   anterior<-paste0("f",k)
# }
# Origem<-c(Origem,anterior)
# Destino<-c(Destino,"end")

#-------------------------------
for (i in 1:n) {
 Origem<-c(Origem,"begin")
 Destino<-c(Destino,paste0("f",i))
}
for (i in 1:n) {
 Origem<-c(Origem,paste0("f",i))
 Destino<-c(Destino,"end")
}
#-------------------------------

iplan_precedencias <- data.frame(Origem,Destino)
write.csv(iplan_precedencias,
          file = "~/Doutorado/iplan-teste-computacional/iplan-precedencias.csv",
          quote = FALSE,
          row.names=FALSE)
###################################################
id<-NULL
beneficioTangivel<-NULL
for (i in 1:n) {
  id<-c(id,paste0("f",i))
  beneficioTangivel<-c(beneficioTangivel,sample(5:200,1))
}
iplan_beneficios_tangiveis <- data.frame(id,beneficioTangivel)
#sum(iplan_beneficios_tangiveis$beneficioTangivel)

#write.csv(iplan_beneficios_tangiveis,
#          file = "~/Doutorado/iplan-teste-computacional/iplan-beneficios-tangiveis.csv",
#          quote = FALSE,
#          row.names=FALSE)
###################################################
id<-"begin"
duracao<-0
investimento<-0
soma<-0
for (i in 1:n) {
  id<-c(id,paste0("f",i))
  duracao<-c(duracao,sample(1:8,1,prob=c(0.05  ,  0.2  ,  0.4  ,  0.6  ,  0.4  ,  0.2  ,  0.05  ,  0.05  )))
  invest<-sample(1:50,1)
  soma<-soma+invest
  investimento<-c(investimento,invest)
}
id<-c(id,"end")
duracao<-c(duracao,"0")
investimento<-c(investimento,"0")
iplan_funcionalidades <- data.frame(id,duracao,investimento)
write.csv(iplan_funcionalidades,
          file = "~/Doutorado/iplan-teste-computacional/iplan-funcionalidades.csv",
          quote = FALSE,
          row.names=FALSE)
####################################################
id<-NULL
duracao<-NULL
for (i in 1:ciclosEntrega) {
  id<-c(id,paste0("CE",i))
  duracao<-c(duracao,duracaoCiclo)
}
iplan_ciclos_entrega <- data.frame(id,duracao)
write.csv(iplan_ciclos_entrega,
          file = "~/Doutorado/iplan-teste-computacional/iplan-ciclos-entrega.csv",
          quote = FALSE,
          row.names=FALSE)
####################################################
 capital<-soma
 iplan_portfolio <- data.frame(capital,juro,juroBeneficioTangivel)
 write.csv(iplan_portfolio,
           file = "~/Doutorado/iplan-teste-computacional/iplan-portfolio.csv",
           quote = FALSE,
           row.names=FALSE)
####################################################
getwd()
setwd("~/Doutorado/iplan-teste-computacional/")
getwd()
edges=read.csv("~/Doutorado/iplan-teste-computacional/iplan-precedencias.csv",comment.char="#")
vertices=read.csv("~/Doutorado/iplan-teste-computacional/iplan-funcionalidades.csv",comment.char="#")
capital=read.csv("~/Doutorado/iplan-teste-computacional/iplan-portfolio.csv",comment.char="#")
releases=read.csv("~/Doutorado/iplan-teste-computacional/iplan-ciclos-entrega.csv",comment.char="#")

l<-enumerateAllPEVs() #Esse método está gerando planos duplicados! Pode ser que tenha processamento desnecessário!
#l<-enumerateAllPEVsV2() #Essa é a versão para usar com o fatorial (toda as combinacoes possiveis de planos de entrega)

l<-sampleValidReleasePlan(2000)

qtdPlanosPopulacao<-length(l)
if(qtdPlanosPopulacao==0){
  loginfo(paste("Nao vamos calcular o DEA pois deu mais de 253.000 planos...",sep = ""))
}else{
  loginfo(paste("Calculando o DEA...",sep = ""))
  pevs<-dataFramePlanosDeEntregaValidos(l)
  xSample<-cbind(pevs$investimento)
  ySample<-cbind(pevs$beneficiosTangiveis,pevs$beneficiosIntangiveis)
  #b1<-1/Benchmarking::dea(xSample,ySample,ORIENTATION="out",RTS="vrs")$eff
  b1<-FEAR::dea(t(xSample),t(ySample),RTS=1,ORIENTATION=2)
  ####b1<-FEAR::boot.sw98(t(xSample),t(ySample),NREP=1,RTS=1,ORIENTATION=2,alpha=0.1,OUTPUT.FARRELL = F)
  pevs1<-cbind(pevs,eficiencia=b1)
  result1<-pevs1[with(pevs1, order(-eficiencia)), ]
  fim.time <- Sys.time()
  time.taken <- fim.time - inicio.time
  write.csv(result1, file = paste(getwd(),"/avaliacao-usando-dea-ID",ID_execucao,"-",n,"-funcionalidades-",inicio.time,".csv",sep=""))
  inicio.time
  fim.time
  loginfo(paste("############ O TEMPO para ",n," FUNCIONALIDADES (PLANOS ",qtdPlanosPopulacao,") FOI ",time.taken," ###############",sep = ""))
}

#save(pevs,file=paste(getwd(),"/pevs-ID",ID_execucao,".Rdata",sep=""))
#save(result1,file=paste(getwd(),"/result1-ID",ID_execucao,".Rdata",sep=""))

###########################################################################
#   BOOSTRAP
###########################################################################

#load(paste(getwd(),"/pevs-ID",ID_execucao,".Rdata",sep=""))
#load(paste(getwd(),"/result1-ID",ID_execucao,".Rdata",sep=""))

qtdPlanosPopulacao<-nrow(pevs)
eficienciaMelhor<-0
limiteInferior<-0
sintoniaPercent<-0
alpha = 0.05     #Testar com 0.003  99,4%
#while(!(sintoniaPercent>=90)){ #!(eficienciaMelhor>limiteInferior)){
for (tamanhoDaAmostra in c(2000)){
  inicio.time <- Sys.time()
  amostraIDs<-sample(seq(1,nrow(pevs),1),tamanhoDaAmostra)
  pevs2<-pevs[amostraIDs,]
  #pevs2<-pevs
  xSample2<-cbind(pevs2$investimento)
  ySample2<-cbind(pevs2$beneficiosTangiveis,pevs2$beneficiosIntangiveis)
  simple<-F
  if(simple){
    bootstrapSamples=500
    loginfo(paste("############ Calculando Boostrap simplicado com amostra de ",tamanhoDaAmostra,", alpha de ",alpha," e ",bootstrapSamples," bootstraps ###############",sep = ""))
    tab2<-behr1(xSample2,ySample2,RTS="VRS",ORIENTATION = "out",NREP=bootstrapSamples)
    loginfo(paste("Terminou o boot.sw98",sep = ""))
    eficienciaSW98<-tab2[4,]
    icInferiorSW98<-tab2[5,]
    icSuperiorSW98<-tab2[6,]
    result2<-cbind(pevs2,eficienciaSW98,icSuperiorSW98,icInferiorSW98)
    k<-result2[with(result2, order(-eficienciaSW98)), ]
    k$index <- as.numeric(row.names(k))
    kk<-k[order(k$index), ]
    pevs2<-cbind(pevs2,eficienciaSW98,icSuperiorSW98,icInferiorSW98)
    result2<-pevs2[with(pevs2, order(-eficienciaSW98)), ]
    maior<-result2[1,]$ciclos
    maior
    rbind(icSuperiorSW98=result2[1,]$icSuperiorSW98,
          icInferiorSW98=result2[1,]$icInferiorSW98,
          eficiencia=subset(result1,ciclos==maior)$eficiencia
    )
    loginfo(paste("[",result2[1,]$icSuperiorSW98,",",result2[1,]$icInferiorSW98,"] ",subset(result1,ciclos==maior)$eficiencia,sep = ""))
    loginfo(paste("Tamanho do intervalo de confianca simplificado: ",(result2[1,]$icSuperiorSW98-result2[1,]$icInferiorSW98)*100,"%",sep = ""))
    fim.time <- Sys.time()
    time.taken <- fim.time - inicio.time
    loginfo(paste("############ O TEMPO para calcular o Boostrap sw98 com ",tamanhoDaAmostra," amostras FOI ",time.taken," ###############",sep = ""))
  }else{
    bootstrapSamples=2000
    loginfo(paste("############ Calculando Boostrap sw98 com amostra de ",tamanhoDaAmostra,", alpha de ",alpha," e ",bootstrapSamples," bootstraps ###############",sep = ""))
    b2<-FEAR::boot.sw98(t(xSample2),t(ySample2),NREP=bootstrapSamples,RTS=1,ORIENTATION=2,OUTPUT.FARRELL = T,CI.TYPE=2,alpha=alpha)
    fim.time <- Sys.time()
    time.taken <- fim.time - inicio.time
    loginfo(paste("Terminou o boot.sw98",sep = ""))
    tab2 <- round(rbind(dhat=b2$dhat,
                        dhat.m=rowMeans(b2$boot),
                        bias=b2$bias,
                        dhat.bc=b2$dhat.bc,
                        ci.low=b2$conf.int[,1],
                        ci.high=b2$conf.int[,2])
                  ,10)
    eficienciaSW98<-1/tab2[4,]
    icInferiorSW98<-1/tab2[6,]
    icSuperiorSW98<-1/tab2[5,]
    result2<-cbind(pevs2,eficienciaSW98,icSuperiorSW98,icInferiorSW98)
    k<-result2[with(result2, order(-eficienciaSW98)), ]
    k$index <- as.numeric(row.names(k))
    kk<-k[order(k$index), ]
    pevs2<-cbind(pevs2,eficienciaSW98,icSuperiorSW98,icInferiorSW98)
    result2<-pevs2[with(pevs2, order(-eficienciaSW98)), ]
    #Analise do melhor
    maior<-result2[1,]$ciclos
    loginfo(paste("Melhor plano: [",result2[1,]$icSuperiorSW98,",",result2[1,]$icInferiorSW98,"] ",subset(result1,ciclos==maior)$eficiencia,sep = ""))
    loginfo(paste("Tamanho do ic para esse melhor: ",(result2[1,]$icSuperiorSW98-result2[1,]$icInferiorSW98)*100,"%",sep = ""))
    eficienciaMelhor<-subset(result1,ciclos==maior)$eficiencia
    limiteInferior<-result2[1,]$icInferiorSW98
    #KOLMOGOROV
    qtdMenor<-nrow(subset(result1,eficiencia<eficienciaMelhor))
    posicaoPercentual<-qtdMenor/qtdPlanosPopulacao
    loginfo(paste("KOLMOGOROV: ",posicaoPercentual*100,"% dos planos na populacao sao menos eficientes que o melhor da amostra bootstrap.",sep = ""))
    #Analise dos "melhores"
    corte<-0.95
    xPlanosBootstrap<-subset(result2,eficienciaSW98>corte)
    loginfo(paste("Existe ",nrow(xPlanosBootstrap)," planos com eficiencia maior do que ",corte,".",sep = ""))
    bstp<-merge(subset(result1,ciclos %in% xPlanosBootstrap$ciclos),
                xPlanosBootstrap[,c(1,5,6,7)],
                by="ciclos")
    bstp<-bstp[with(bstp, order(-eficienciaSW98)), ]
    bstp<-cbind(bstp,isIcOK=bstp$icInferiorSW98<=bstp$eficiencia)
    t<-table(bstp$icInferiorSW98<=bstp$eficiencia)
    sintonia<-t[names(t)==TRUE]
    sintoniaPercent<-(sintonia/nrow(xPlanosBootstrap))*100
    loginfo(paste("Desses ",nrow(xPlanosBootstrap)," planos, ",sintonia," (",sintoniaPercent,"%) tiveram o ic em sintonia com a eficiencia da populacao.",sep = ""))
    #save(bstp,file=paste(getwd(),"/bstp-ID",ID_execucao,".Rdata",sep=""))
    #write.csv2(bstp, file = paste(getwd(),"/avaliacao-usando-dea-ID",ID_execucao,"-bootstrap---com-eficiencia-real.csv",sep=""), row.names=FALSE, quote = FALSE,na="")
    #
    loginfo(paste("############ O tempo para calcular com ",tamanhoDaAmostra," amostras foi ",time.taken," ###############",sep = ""))
  }
  #if(sintoniaPercent>=90) break
}

#save(result2,file=paste(getwd(),"/result2-bootstrap-sw98-ID",ID_execucao,".Rdata",sep=""))
#write.csv(result2, file = paste(getwd(),"/avaliacao-usando-dea-ID",ID_execucao,"-bootstrap.csv",sep=""))

head(bstp,6)

#ID_execucao<-12

#load(paste(getwd(),"/bstp-ID",ID_execucao,".Rdata",sep=""))
#load(paste(getwd(),"/result2-bootstrap-sw98-ID",ID_execucao,".Rdata",sep=""))

###########################
precedencia=read.csv("~/Doutorado/iplan-teste-computacional/iplan-precedencias.csv",comment.char="#")
colnames(precedencia) <- c("predecessor", "id")
precedencia$predecessor[precedencia$predecessor=="begin"]<-NA
#precedencia$sucessor<-substring(precedencia$sucessor, 2)
tangiveis=read.csv(paste(getwd(),"/iplan-beneficios-tangiveis.csv",sep=""),comment.char="#")
tangiveis$beneficioTangivel<-tangiveis$beneficioTangivel*0.1
duracaoEinvestimento=head(vertices[-c(1),],-1)
duracaoEinvestimento$investimento<-duracaoEinvestimento$investimento*0.1
intangiveis=read.csv(paste(getwd(),"/AHP-resultado-intangiveis.csv",sep=""),comment.char="#")
colnames(intangiveis) <- c("id", "beneficioIntangivel")
intangiveis$beneficioIntangivel<-sapply(intangiveis$beneficioIntangivel*100, round, 2)
df<-merge(duracaoEinvestimento,tangiveis,by = "id")
df<-merge(df,intangiveis,by="id") 
df<-merge(df,precedencia,by="id") 
df$ordem<-substring(df$id, 2)
df[, c(7)] <- sapply(df[, c(7)], as.numeric)
df<-df[order(df$ordem),]
df$ordem<-NULL
df
write.csv2(df, file = paste(getwd(),"/dados-funcionalidades.csv",sep=""), row.names=FALSE, quote = FALSE,na="")


planos<-head(result1,37)
library(stringr)
p<-str_split_fixed(planos$ciclos, ".  CE2: ", 2)
ce1<-substring(p[,1],13)
ce2<-str_split_fixed(p[,2], "\\.", 2)
ce2<-ce2[,1]
ce<-data.frame(id=rownames(planos),um=ce1,dois=ce2)
write.csv2(ce, file = paste(getwd(),"/dados-planos.csv",sep=""), row.names=FALSE, quote = FALSE,na="")

#==============================

planos<-xPlanosBootstrap
library(stringr)
p<-str_split_fixed(planos$ciclos, ".  CE2: ", 2)
ce1<-substring(p[,1],13)
ce2<-str_split_fixed(p[,2], "\\, end.", 2)
ce2<-ce2[,1]
ce<-data.frame(id=rownames(planos),um=ce1,dois=ce2,
               invest=xPlanosBootstrap$investimento,
               tang=xPlanosBootstrap$beneficiosTangiveis,
              intang=xPlanosBootstrap$beneficiosIntangiveis,
               effMedia=format(round(xPlanosBootstrap$eficienciaSW98*100,1), digits=3, decimal.mark=","),
              icSup=format(round(xPlanosBootstrap$icSuperiorSW98*100,1), digits=3, decimal.mark=","),
              icInf=format(round(xPlanosBootstrap$icInferiorSW98*100,1), digits=3, decimal.mark=","))
write.csv2(ce, file = paste(getwd(),"/dados-planos-bootstrap.csv",sep=""), row.names=FALSE, quote = FALSE,na="")


#####################

behr1<-function(x,y,RTS="VRS",ORIENTATION = "in",NREP=1000){
  library(Benchmarking)
  n<-length(x)
  dhat<-1/dea(x,y,RTS = RTS,ORIENTATION = ORIENTATION)$eff #O original é 1/dea
  #dea.plot(x,y,RTS = RTS,txt=1:4)
  B<-NREP
  Bm<-matrix(NA,B,n)
  for(i in 1:B){
    dstar<-sample(dhat,n,replace=T)
    #dstar<-c(1, 0.3888889, 0.3676768, 0.3888889)
    #dstar<-c(1,2.571429,2.719780,2.571429)
    if(ORIENTATION == "in"){
      xs<-x*dstar/dhat
      #dea.plot(c(x,xs),c(y,y), RTS="vrs",txt=1:8)
      #dea.plot(xs,y, RTS="vrs",txt=5:8)
      #dea.plot(c(xs,20*0.1512346),c(y,18.9), RTS="vrs",txt=5:9)
      Bm[i,]<-1/dea(x,y,XREF=xs,YREF = y,RTS = RTS,ORIENTATION = ORIENTATION)$eff #O original é 1/dea
    }else{
      ys<-y*(dstar/dhat)
      #dea.plot(c(x,xs),c(y,y), RTS="vrs",txt=1:8)
      #dea.plot(xs,y, RTS="vrs",txt=5:8)
      #dea.plot(c(xs,20*0.1512346),c(y,18.9), RTS="vrs",txt=5:9)
      Bm[i,]<-1/dea(x,y,XREF=x,YREF = ys,RTS = RTS,ORIENTATION = ORIENTATION)$eff #O original é 1/dea
    }
  }
  dhat.m<-colMeans(Bm)
  dhat.bias<-dhat.m-dhat
  dhat.bc<-dhat-dhat.bias
  #Bm.c <- t(apply(-Bm,1,function(z) z+2*dhat)) 
  ci.low<-apply(Bm, 2, quantile,0.005)-2*dhat.bias
  #ci.low <- apply(Bm.c,2,quantile,0.05) 
  ci.high<-apply(Bm, 2, quantile,0.995)-2*dhat.bias
  #ci.high <- apply(Bm.c,2,quantile,0.95)
  tab<-round(rbind(dhat,dhat.m,dhat.bias,dhat.bc,ci.low,ci.high),3)
  tab
}