#######################################vision optimiste
sleep<-1-as.matrix(is.na(experts))
agg.online<- mixture(Y = data1$REALISE_RE , experts = experts, model = 'MLpol', loss.type = 'square',
loss.gradient = TRUE,awake=sleep)
plot(agg.online)
summary(agg.online)
rmse(data1$REALISE_RE, agg.online$prediction)
#######################################vision "pessismiste"
sleep<-1-as.matrix(is.na(experts))
agg.onlineCAPP<- mixture(Y = data1$SIGNAL_CAPP , experts = experts, model = 'MLpol', loss.type = 'square',
loss.gradient = TRUE,awake=sleep)
plot(agg.onlineCAPP)
rmse(data1$REALISE_RE, agg.onlineCAPP$prediction)
#######################analyse des résultats
rmse(data1$REALISE_RE, agg.online$prediction)
rmse(data1$REALISE_RE, agg.onlineCAPP$prediction)
rmse(data1$REALISE_RE,data1$CPODEPIL_PU_J1)
rmse(data1$REALISE_RE,data1$DCO_PU_MOY_J1)
#save.image(file = "C:\\Havana\\POC_doaat\\Results\\test1.RData")
########################################################################
########################aggregation réaliste
########################################################################
l<-3*7*48
agg.forecast<-array(0,dim=nrow(data1))
agg.forecast[1]<-mean(experts[1,]*sleep)
for(i in c(1:nrow(data1)))
{
if(i<=l)
{
Y<-data1$SIGNAL_CAPP[1:i]
}
if(i>l)
{
Y<-c(data1$REALISE_RE[1:(i-l)], data1$SIGNAL_CAPP[(i-l+1):i])
}
agg<- mixture(Y, experts = experts[1:i,], model = 'MLpol', loss.type = 'square',
loss.gradient = TRUE,awake=sleep[1:i,])
agg.forecast[i+1]<-predict(agg, newexpert = experts[(i+1),], online = F,type="response")
}
rmse(data1$REALISE_RE,agg.forecast)
plot(agg)
plot(agg)
summary(agg)
plot(Y,type='l')
lines(data1$REALISE_RE,col='red')
i
l
########################aggregation réaliste
########################################################################
l<-3*7
agg.forecast<-array(0,dim=nrow(data1))
agg.forecast[1]<-mean(experts[1,]*sleep)
for(i in c(1:nrow(data1)))
{
if(i<=l)
{
Y<-data1$SIGNAL_CAPP[1:i]
}
if(i>l)
{
Y<-c(data1$REALISE_RE[1:(i-l)], data1$SIGNAL_CAPP[(i-l+1):i])
}
agg<- mixture(Y, experts = experts[1:i,], model = 'MLpol', loss.type = 'square',
loss.gradient = TRUE,awake=sleep[1:i,])
agg.forecast[i+1]<-predict(agg, newexpert = experts[(i+1),], online = F,type="response")
}
rmse(data1$REALISE_RE,agg.forecast)
plot(agg)
summary(agg)
plot(Y,type='l')
lines(data1$REALISE_RE,col='red')
rmse(data1$REALISE_RE,agg.forecast)
modelList<-c("earth","gbm","xgbTree")   #,"svmPoly","kernelpls"
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.8
########################stacking experts
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"CPO","DCO")
dim(experts)
modelList<-c("earth","gbm","ppr")   #,"svmPoly","kernelpls"
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.8
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.8
########################stacking experts
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"CPO","DCO")
modelList<-c("earth","gbm")   #,"svmPoly","kernelpls"
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.8
########################stacking experts
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"CPO","DCO")
warnings()
summary(data1$EJP_NORD_EDF)
plot(data1$EJP_NORD_EDF)
plot(data0$EJP_NORD_EDF)
plot(data0$EJP_PACA_EDF)
plot(data0$TEMPO_BLANC)
plot(data0$TEMPO_ROUGE)
rm(list=objects())
###############packages
library(dygraphs)
library(xts)
library(lubridate)
library(stackeR)
library(forecast)
library(opera)
Data<-readRDS(file="C:\\Havana\\POC_doaat\\Data\\Data_doaat_V5.RDS")
a<-which(Data$Date==strptime("2015-08-31 23:30:00", "%Y-%m-%d %H:%M:%S"))
data0<-Data[1:a,]
data1<-Data[(a+1):nrow(Data),]
summary(data0$CPODEPIL_PU_J1)
summary(data0$DCO_PU_MOY_J1)
sela<-which(data0$Instant==24)
selb<-which(data1$Instant==24)
data0<-data0[sela,]
data1<-data1[selb,]
names(data0)
cov_demand<-c("REALISE_RE.1008","SIGNAL_CAPP.1008","ETR_BRUTE.48","SIGNAL_CAPP.48","SIGNAL_CAPP.1008")
cov_tarif<-c("TEMPO_ROUGE","TEMPO_BLANC","EJP_NORD_EDF","EJP_OUEST_EDF","EJP_SUD_EDF","EJP_PACA_EDF" )
cov_meteo<-c("T_REAL","NEB_REAL","VENT_REAL","T_NORM","NEB_NORM","VENT_NORM")
cov_calendaire<-c("dimanche", "jeudi", "lundi","mardi","mercredi","samedi","vendredi","BH0","BH1","Posan","Trend","ERDF_Rupture11")
cov_model<-c("CPODEPIL_PU_J1","DCO_PU_MOY_J1")
#nom<-c(cov_demand,cov_tarif,cov_meteo,cov_calendaire,cov_model)
nom<-c(cov_demand,cov_meteo,cov_calendaire,cov_model)
Dataa_matrix<-as.matrix(data0[,nom])
Datab_matrix<-as.matrix(data1[,nom])
summary(Dataa_matrix)
#modelList<-c("earth","ppr","gbm","xgbTree")   #,"svmPoly","kernelpls"
modelList<-c("earth","gbm","xgbTree")   #,"svmPoly","kernelpls"
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.8
########################stacking experts
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"CPO","DCO")
dim(experts)
# experts<-cbind(stak$forecast,stak$forecast.stack,prevARIMA,prevETS,prevHW)
# colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"ARIMA","ETS","HW")
#######################################vision optimiste
sleep<-1-as.matrix(is.na(experts))
agg.online<- mixture(Y = data1$REALISE_RE , experts = experts, model = 'MLpol', loss.type = 'square',
loss.gradient = TRUE,awake=sleep)
plot(agg.online)
summary(agg.online)
rmse(data1$REALISE_RE, agg.online$prediction)
cov_calendaire<-c("dimanche", "jeudi", "lundi","mardi","mercredi","samedi","vendredi","BH0","BH1","Posan","Trend")
cov_demand<-c("REALISE_RE.1008","SIGNAL_CAPP.1008","ETR_BRUTE.48","SIGNAL_CAPP.48","SIGNAL_CAPP.1008")
cov_tarif<-c("TEMPO_ROUGE","TEMPO_BLANC","EJP_NORD_EDF","EJP_OUEST_EDF","EJP_SUD_EDF","EJP_PACA_EDF" )
cov_meteo<-c("T_REAL","NEB_REAL","VENT_REAL","T_NORM","NEB_NORM","VENT_NORM")
#cov_calendaire<-c("dimanche", "jeudi", "lundi","mardi","mercredi","samedi","vendredi","BH0","BH1","Posan","Trend","ERDF_Rupture11")
cov_calendaire<-c("dimanche", "jeudi", "lundi","mardi","mercredi","samedi","vendredi","BH0","BH1","Posan","Trend")
cov_model<-c("CPODEPIL_PU_J1","DCO_PU_MOY_J1")
#nom<-c(cov_demand,cov_tarif,cov_meteo,cov_calendaire,cov_model)
nom<-c(cov_demand,cov_meteo,cov_calendaire,cov_model)
Dataa_matrix<-as.matrix(data0[,nom])
Datab_matrix<-as.matrix(data1[,nom])
summary(Dataa_matrix)
#modelList<-c("earth","ppr","gbm","xgbTree")   #,"svmPoly","kernelpls"
modelList<-c("earth","gbm","xgbTree")   #,"svmPoly","kernelpls"
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.8
########################stacking experts
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"CPO","DCO")
dim(experts)
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.5
########################stacking experts
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"CPO","DCO")
dim(Dataa_matrix)
summary(Dataa_matrix)
#modelList<-c("earth","ppr","gbm","xgbTree")   #,"svmPoly","kernelpls"
modelList<-c("earth","gbm","ppr")   #,"svmPoly","kernelpls"
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.5
########################stacking experts
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"CPO","DCO")
#modelList<-c("earth","ppr","gbm","xgbTree")   #,"svmPoly","kernelpls"
modelList<-c("earth","gbm","ppr")   #,"svmPoly","kernelpls"
trControl<-trainControl("repeatedcv", repeats=1,number=5)
train.rate<-0.8
########################stacking experts
stak<-tryCatch(stak<-stacking(x0=Dataa_matrix,x1=Datab_matrix,y0=data0$REALISE_RE,train.rate=train.rate,modelList,trControl,data.export=T,sample.type="random")
, error = function(e){NULL})
experts<-cbind(stak$forecast,stak$forecast.stack,data1$CPODEPIL_PU_J1,data1$DCO_PU_MOY_J1)
colnames(experts)<-c(colnames(stak$forecast),colnames(stak$forecast.stack),"CPO","DCO")
dim(experts)
library(cranlogs)
cran_top_downloads(when='last-day')
?cranlogs
data <- cran_downloads(packages=c("opera"), from="2016-08-17")
names(data)
par(mfrow=c(1,1))
plot(data$date,cumsum(data$count),type='b',pch=20,col='purple',xlab='Date',ylab='Nb of Downloads')
library(cranlogs)
cran_top_downloads(when='last-day')
?cranlogs
data <- cran_downloads(packages=c("opera"), from="2016-08-17")
names(data)
par(mfrow=c(1,1))
plot(data$date,cumsum(data$count),type='b',pch=20,col='purple',xlab='Date',ylab='Nb of Downloads')
data <- cran_downloads(packages=c("mgcv","ggplot2","randomForest","gbm","caret"), from="2013-01-01")
names(data)
par(mfrow=c(1,1))
sel<-which(data$package=="mgcv")
plot(data$date[sel],cumsum(data$count[sel]),type='l',pch=20,col='purple',ylim=c(0,10*10^6))
s
data <- cran_downloads(packages=c("mgcv","ggplot2","randomForest","gbm","caret"), from="2013-01-01")
data <- cran_downloads(packages=c("mgcv","ggplot2","randomForest","gbm","caret"), from="2013-01-01")
data <- cran_downloads(packages=c("mgcv","ggplot2","randomForest","gbm","caret"), from="2013-01-01")
data <- cran_downloads(packages=c("mgcv","ggplot2","randomForest","gbm","caret"), from="2013-01-01")
data <- cran_downloads(packages=c("mgcv","ggplot2","randomForest","gbm","caret"), from="2013-01-01")
plot(data$date[sel],cumsum(data$count[sel]),type='l',pch=20,col='purple',ylim=c(0,10*10^6))
sel<-which(data$package=="mgcv")
plot(data$date[sel],cumsum(data$count[sel]),type='l',pch=20,col='purple',ylim=c(0,10*10^6))
plot(data$date[sel],cumsum(data$count[sel])
plot(data$date[sel],cumsum(data$count[sel]),type='l',pch=20,col='purple')
par(mfrow=c(1,1))
sel<-which(data$package=="mgcv")
plot(data$date[sel],cumsum(data$count[sel]),type='l',pch=20,col='purple',ylim=c(0,10*10^6))
sel<-which(data$package=="ggplot2")
lines(data$date[sel],cumsum(data$count[sel]),col='blue')
sel<-which(data$package=="randomForest")
lines(data$date[sel],cumsum(data$count[sel]),col='red')
sel<-which(data$package=="gbm")
lines(data$date[sel],cumsum(data$count[sel]),col='green')
sel<-which(data$package=="caret")
lines(data$date[sel],cumsum(data$count[sel]),col='pink')
legend("top",c("mgcv","ggplot2","randomForest","gbm","caret"),col=c("purple","blue","red","green","pink"),lty=1)
par(mfrow=c(1,1))
sel<-which(data$package=="mgcv")
plot(data$date[sel],cumsum(data$count[sel]),type='l',pch=20,col='purple',ylim=c(0,8*10^6))
sel<-which(data$package=="ggplot2")
lines(data$date[sel],cumsum(data$count[sel]),col='blue')
sel<-which(data$package=="randomForest")
lines(data$date[sel],cumsum(data$count[sel]),col='red')
sel<-which(data$package=="gbm")
lines(data$date[sel],cumsum(data$count[sel]),col='green')
sel<-which(data$package=="caret")
lines(data$date[sel],cumsum(data$count[sel]),col='pink')
legend("top",c("mgcv","ggplot2","randomForest","gbm","caret"),col=c("purple","blue","red","green","pink"),lty=1)
par(mfrow=c(1,1))
sel<-which(data$package=="mgcv")
plot(data$date[sel],cumsum(data$count[sel]),type='l',pch=20,col='purple',ylim=c(0,6*10^6))
sel<-which(data$package=="ggplot2")
lines(data$date[sel],cumsum(data$count[sel]),col='blue')
sel<-which(data$package=="randomForest")
lines(data$date[sel],cumsum(data$count[sel]),col='red')
sel<-which(data$package=="gbm")
lines(data$date[sel],cumsum(data$count[sel]),col='green')
sel<-which(data$package=="caret")
lines(data$date[sel],cumsum(data$count[sel]),col='pink')
legend("top",c("mgcv","ggplot2","randomForest","gbm","caret"),col=c("purple","blue","red","green","pink"),lty=1)
sel<-which(data$package=="mgcv")
cumsum(data$count[sel])
tail(cumsum(data$count[sel]))
par(mfrow=c(1,1))
sel<-which(data$package=="mgcv")
plot(data$date[sel],cumsum(data$count[sel]),type='l',pch=20,col='purple',ylim=c(0,6*10^6),xlab='Date',ylab='Downloads')
sel<-which(data$package=="ggplot2")
lines(data$date[sel],cumsum(data$count[sel]),col='blue')
sel<-which(data$package=="randomForest")
lines(data$date[sel],cumsum(data$count[sel]),col='red')
sel<-which(data$package=="gbm")
lines(data$date[sel],cumsum(data$count[sel]),col='green')
sel<-which(data$package=="caret")
lines(data$date[sel],cumsum(data$count[sel]),col='pink')
legend("top",c("mgcv","ggplot2","randomForest","gbm","caret"),col=c("purple","blue","red","green","pink"),lty=1)
271+1437+3600
271+1437+366
271+1437+366+900
1500+665+795-50
1500+665+700
library(opera)
?opera
?mixture
plot(lambda/n,type='l')
lines(exp(-lambda/n)-1,col='red')
lambda<-0.5
n<-c(10:1000)
plot(lambda/n,type='l')
lines(exp(-lambda/n)-1,col='red')
plot(lambda/n,type='l')
lines(exp(lambda/n)-1,col='red')
cr<-lambda/n
var<-exp(lambda/n)-1
plot(lambda/n,type='l')
lines(var,col='red')
plot(cr-var)
log(5/100)
n<-c(1:100)
lamba0<-1
lamba1<-5
alpha<-5/100
beta<-10/100
ka<-n*lambda0*log(1/(1-alpha))
kb<--n*lamba1*log(beta)
n<-c(1:100)
lambda0<-1
lambda1<-5
alpha<-5/100
beta<-10/100
ka<-n*lambda0*log(1/(1-alpha))
kb<--n*lambda1*log(beta)
plot(ka,type='l')
lines(kb,col='red')
plot(ka,type='l',ylim=range(ka,kb))
lines(kb,col='red')
n<-c(1:100)
lambda0<-1
lambda1<-5
alpha<-5/100
beta<-5/100
ka<-n*lambda0*log(1/(1-alpha))
kb<--n*lambda1*log(beta)
plot(ka,type='l',ylim=range(ka,kb))
lines(kb,col='red')
?Chisquare
2*114 +0.7*165
n<-1000
y<-rnorm(n)
hist(y,breaks=100)
h<-hist(y,breaks=100)
names(h)
h$density
lines(h$density,type='l'
)
?hist
plot(h$density,type='b')
sum(h$density)
h$mids
diff(h$mids)
sum(h$density*diff(h$mids)[1])
?density
h<-hist(y,breaks=100,freq=T)
h<-hist(y,breaks=100,freq=F)
lines(h$density,type='l')
lines(h$mids,h$density,type='l')
h<-hist(y,breaks=100,freq=F)
lines(h$mids,h$density,type='l')
d<-density(y,kernel = c("gaussian"),bw=0.5)
d<-density(y,kernel = c("gaussian"),bw=0.5)
h<-hist(y,breaks=100,freq=F)
lines(h$mids,d,type='l')
length(d)
names(d)
h<-hist(y,breaks=100,freq=F)
lines(d$x,d$y,type='l',col='red')
d<-density(y,kernel = c("gaussian"),bw=1)
h<-hist(y,breaks=100,freq=F)
lines(d$x,d$y,type='l',col='red')
d<-density(y,kernel = c("gaussian"),bw=0.1)
h<-hist(y,breaks=100,freq=F)
lines(d$x,d$y,type='l',col='red')
rm(list=objects())
library(opera)
library(stackeR)
library(magrittr)
library(qgam)
?qgam
144*5/60
?gc
gc()
install.packages("feather")
rm(list=objects())
library(feather)
data<-feather("data/SALELH_NOEUDS_PASSAGE2.feather")
80+970+120
library(cranlogs)
cran_top_downloads(when='last-day')
data <- cran_downloads(packages=c("mgcv","ggplot2","randomForest","gbm","caret"), from="2013-01-01")
cran_top_downloads(when='last-day')
cran_top_downloads(when='last-day')
?cranlogs
library(cranlogs)
data <- cran_downloads(packages=c("opera"), from="2016-08-17")
names(data)
par(mfrow=c(1,1))
plot(data$date,cumsum(data$count),type='b',pch=20,col='purple',xlab='Date',ylab='Nb of Downloads')
<
data <- cran_downloads(packages=c("opera"), from="2016-08-17")
data <- cran_downloads(packages=c("opera"), from="2016-08-17")
data <- cran_downloads(packages=c("opera"), from="2016-08-17")
data <- cran_downloads(packages=c("opera"), from="2016-08-17")
data <- cran_downloads(packages=c("opera"), from="2016-08-17")
names(data)
par(mfrow=c(1,1))
plot(data$date,cumsum(data$count),type='b',pch=20,col='purple',xlab='Date',ylab='Nb of Downloads')
rm(list=objects())
######Import des donnees
setwd("C:\\Enseignement\\2016-2017\\Serie_temp\\TP\\TP1_data\\")
beer<-read.csv("beer2.csv",header=TRUE,skip=1)
head(beer)
summary(beer)
plot(beer$BeerProd,type='b',pch=20)
#######creation de la date
date1<- strptime(c("01/01/91"), "%m/%d/%y")
date2<- strptime(c("08/01/95"), "%m/%d/%y")
Date<-seq(date1,date2,by = "1 month")
beer<-data.frame(Date,beer$BeerProd)
names(beer)<-c("Date","BeerProd")
summary(beer)
plot(beer$Date,beer$BeerProd,type='l')
#########classe ts
beer.ts<-ts(beer$BeerProd,start=1,frequency=12)
plot(beer.ts)
#########classe zoo
library(zoo)
beer.zoo<-zoo(beer$BeerProd,order.by=beer$Date)
plot(beer.zoo)
######################statistiques de base
mean(beer$BeerProd)
sd(beer$BeerProd)
summary(beer)
boxplot(beer$BeerProd)
hist(beer$BeerProd,breaks=20)
year<-format(beer$Date,"%y")
mean.year<-tapply(beer$BeerProd,as.factor(year),mean)
plot(mean.year,type='b',axes=F)
axis(1,c(1:5),names(mean.year))
axis(2)
month<-format(beer$Date,"%m")
mean.month<-tapply(beer$BeerProd,as.factor(month),mean)
plot(mean.month,type='b',axes=F)
axis(1,c(1:16),names(mean.month))
axis(2)
mean.month
month<-format(beer$Date,"%m")
mean.month<-tapply(beer$BeerProd,as.factor(month),mean)
plot(mean.month,type='b',axes=F)
axis(1,c(1:12),names(mean.month))
axis(2)
year<-format(beer$Date,"%Y")
mean.year<-tapply(beer$BeerProd,as.factor(year),mean)
plot(mean.year,type='b',axes=F)
axis(1,c(1:5),names(mean.year))
axis(2)
data<-read.table("conso_2015.csv",header=T,sep=';')
head(data) ###les données sont par jour puis par heure, on préfèrera une date incluant l'heure
data<-read.table("conso_2015.csv",header=T,sep=';')
head(data) ###les données sont par jour puis par heure, on préfèrera une date incluant l'heure
date1<- strptime("01/01/2015 00:00:00", "%m/%d/%Y %H:%M:%S")
date2<- strptime("11/30/2015 23:30:00", "%m/%d/%Y %H:%M:%S")
Date<-seq.POSIXt(date1,date2,by = "30 min")
summary(Date)
head(Date)
X<-as.matrix(t(data[,-1]))
conso<-c(X)
head(conso,50)
plot(conso[1:(48*7)],type='l')
plot(conso[1:(48*7)],type='l')
plot(Date,conso,type='l')
library(xts)
plot(Date,conso,type='l')
library(xts)
conso.xts<-xts(conso,order.by=Date)
plot(conso.xts)
mean(conso.xts)
month<-as.factor(.indexmon(conso.xts))
mean.month<-tapply(conso.xts,month,mean)
plot(mean.month,type='b')
dow<-as.factor(.indexwday(conso.xts))
tapply(conso.xts,dow,mean)
plot(conso.xts,type='b')
conso_day<-tapply(conso.xts,dow,mean)
plot(conso_day,type='b')
hour<-as.factor(.indexhour(conso.xts))
mean.dow.hour<-tapply(conso.xts,dow:hour,mean)
plot(mean.dow.hour,type='l')
abline(v=seq(1,24*7,by=24))
plot(mean.dow.hour,type='l')
abline(v=seq(1,24*7,by=24),col='red')
col.pal<-colorRampPalette(c("lightblue", "red"))( 12 )
sel<-which(.indexhour(conso.xts)==20)
boxplot(conso[sel]~month[sel],col=col.pal)
col.pal<-colorRampPalette(c("lightblue", "red"))( 12 )
sel<-which(.indexhour(conso.xts)==20)
boxplot(conso[sel]~month[sel],col=col.pal)
col.pal<-colorRampPalette(c("lightblue", "red"))( 12 )
sel<-which(.indexhour(conso.xts)==20)
boxplot(conso[sel]~month[sel],col=col.pal)
names(data)
acf(conso,lag.max=7*48*2)
par(mfrow=c(2,1))
acf(conso,lag.max=48)
par(mfrow=c(2,1))
acf(conso,lag.max=48)
acf(conso,lag.max=7*48*2)
par(mfrow=c(1,2))
acf(conso,lag.max=48)
acf(conso,lag.max=7*48*2)
?acf
