#####################################
#Set up
	rm(list=ls(all=TRUE))
	library(ggplot2)
	library(quantreg)
	theme_set(theme_bw())
	setwd("I:/Project/2017 LD GHG Rule/Footprint_analysis/Outputs")
	fleetorig <- read.table("I:/Project/2017 LD GHG Rule/Footprint_analysis/fleet/20110520_fleet.txt", header=TRUE, sep="\t", na.strings="NA", dec=".", strip.white=TRUE)

#Drop Hybrids, Electrics and Diesels
fleet<-fleetorig[fleetorig$Engine.Cycle %in% c("G","R"),]

##Calculate Weights - Don't remove vehicles after this, as these are based on total volumes
	a<-subset(fleet,NHTSA.Defined.New.NHTSA.Car.Truck == "C")
	car.vol<-sum(a$Volume.2021)
	b<-subset(fleet,NHTSA.Defined.New.NHTSA.Car.Truck == "T")
	truck.vol<-sum(b$Volume.2021)
	fleet$Weight<-ifelse(fleet$NHTSA.Defined.New.NHTSA.Car.Truck=="C",fleet$Volume.2021/car.vol,fleet$Volume.2021/truck.vol)
	
annot<-function(a)
	{paste(signif(coef(a)[[1]],digits=3),"+",signif(coef(a)[[2]],digits=3),"x :R2=",signif(summary(a)[[8]],digits=3))}
	
annotrq<-function(a)
	{paste(signif(coef(a)[[1]],digits=3),"+",signif(coef(a)[[2]],digits=3),"x")}
	
analysis<-function(x,col,title) {
	fleet1<-subset(fleet,NHTSA.Defined.New.NHTSA.Car.Truck == x)
	CO2<-cbind(fleet1$No.Tech.CO2,fleet1$CO2.Package.1,fleet1$CO2.Package.2)
	lm.1<-lm(fleet1$Horsepower/fleet1$Curb.Weight ~ fleet1$Footprint..PU.Average, weights=fleet1$Weight)
	lm.2<-lm(fleet1$Horsepower/fleet1$Curb.Weight ~ poly(fleet1$Footprint..PU.Average,2,raw=TRUE), weights=fleet1$Weight)
	lm.3<-lm(fleet1$Horsepower~poly(fleet1$Curb.Weight,1, raw=TRUE), weights=fleet1$Weight)
	lm.4<-lm(fleet1$Horsepower~poly(fleet1$Curb.Weight,2, raw=TRUE), weights=fleet1$Weight)
	lm.5<-lm(CO2[,col] ~ fleet1$Footprint..PU.Average, weights=fleet1$Weight)

	
#Rick's method for adjusting hp.wt
	lm.8<-lm(CO2[,col] ~ I(fleet1$Horsepower/fleet1$Curb.Weight)+I(fleet1$Curb.Weight),weights = fleet1$Weight)
	#Remove HP.Wt Adjustment, 
	#fleet1$hp.wt.avg<-sum(fleet1$Horsepower/fleet1$Curb.Weight*fleet1$Weight)
	fleet1$hp.wt.CO2<-CO2[,col]#-(fleet1$Horsepower/fleet1$Curb.Weight-fleet1$hp.wt.avg)*coef(lm.8)[2])
	lm.9<-lm(fleet1$hp.wt.CO2 ~ fleet1$Footprint..PU.Average, weights=fleet1$Weight)
	lm.10<-lm(CO2[,col] ~ fleet1$Footprint..PU.Average + I(fleet1$Horsepower/fleet1$Curb.Weight)+I(fleet1$Curb.Weight/fleet1$Footprint..PU.Average),weights = fleet1$Weight)
	
#Adjustment for HP.WT and FP.WT
	fleet1$crb.wt.fp.avg<-sum(fleet1$Curb.Weight/fleet1$Footprint..PU.Average*fleet1$Weight)
	fleet1$hpwt.crbwt.norm.CO2<-fleet1$hp.wt.CO2-coef(lm.8)[3]*(fleet1$Curb.Weight-fleet1$crb.wt.fp.avg*fleet1$Footprint..PU.Average)
	lm.11<-lm(fleet1$hpwt.crbwt.norm.CO2~ fleet1$Footprint..PU.Average, weights=fleet1$Weight)

#FP/WT v. FP.
	lm.12<-lm(I(fleet1$Curb.Weight/fleet1$Footprint..PU.Average) ~ fleet1$Footprint..PU.Average, weights = fleet1$Weight)
	
#CO2 v FP + HP/WT + WT/FP	
	lm.13<-lm(CO2[,col] ~ I(fleet1$Curb.Weight/fleet1$Footprint..PU.Average) + fleet1$Footprint..PU.Average + I(fleet1$Horsepower/fleet1$Curb.Weight), weights = fleet1$Weight)
	lm.14<-lm(CO2[,col] ~ I(fleet1$Curb.Weight/fleet1$Footprint..PU.Average) + poly(fleet1$Footprint..PU.Average,2,raw=TRUE) + I(fleet1$Horsepower/fleet1$Curb.Weight), weights = fleet1$Weight)
	lm.15<-lm(CO2[,col] ~ I(fleet1$Curb.Weight/fleet1$Footprint..PU.Average) + poly(fleet1$Footprint..PU.Average,3,raw=TRUE) + I(fleet1$Horsepower/fleet1$Curb.Weight), weights = fleet1$Weight)
	
			
			
	ans<-list(fleet1,lm.1,lm.2,lm.3,lm.4,lm.5,lm.8,lm.9,lm.10,lm.11,lm.12,lm.13,lm.14,lm.15)
	invisible(return(ans))
}

		
mregres<-function(vtype,a){	
fleeta<-subset(fleet4,NHTSA.Defined.New.NHTSA.Car.Truck == vtype)
#####WT/FP
olsa1<-lm( I(fleeta$Curb.Weight/fleeta$Footprint..PU.Average)~ fleeta$Footprint..PU.Average,weights = fleeta$Weight)

#####HP/WT
olsb1<-lm( I(fleeta$Horsepower/fleeta$Curb.Weight)~ fleeta$Footprint..PU.Average,weights = fleeta$Weight)


return(list(olsa1,olsb1))
}

dchart<-function(dtt,tit,hj){ 
#Unadjusted CO2

dtt$art<-fleet4$CO2.Package.1

clabels<-list(annot(cars.no.tech[[6]]),annot(cars.egr[[6]]))
tlabels<-list(annot(trucks.no.tech[[6]]),annot(trucks.egr[[6]]))

#Base CO2 - Weighted OLS (faceted B& W)
base.chart<-ggplot(data=dtt, aes(y=art, x=Footprint..PU.Average,  size=Volume.2021, weight=Weight))+
		geom_point(aes(shape=1))+
		labs(x="Footprint", y="Technology Normalized CO2", size="2021 Projected Sales")+
		opts(title=paste(tit,"CO2 v. Footprint"))+
		facet_wrap(~NHTSA.Defined.New.NHTSA.Car.Truck)+
		scale_y_continuous(limits=c(0,500))+
		scale_x_continuous(limits=c(35,80))



#Base CO2 - Weighted OLS

base.ols<-ggplot(data=dtt, aes(y=art, x=Footprint..PU.Average, color=NHTSA.Defined.New.NHTSA.Car.Truck, size=Volume.2021, weight=Weight))+
		geom_point(aes(shape=1))+
		geom_smooth(method=lm, formula= y~ poly(x,1, raw=TRUE), se=FALSE)+
		labs(x="Footprint", y="Base CO2", colour="C/T")+
		opts(title=paste(tit,"Base CO2 v. Footprint (Wt. OLS)"))+
		coord_cartesian(xlim = c(35,80), ylim=c(0,500))+
		annotate(geom="text",x=50,y=50, colour="black", label= paste("Car:",clabels[[hj]]) , size=3 ) +
		annotate(geom="text", x = 50, y= 100, colour="black", label=paste("Truck:",tlabels[[hj]]) , size=3 ) 


#WT/FP Normalized CO2  - OLS
clabels<-list(annot(cars.no.tech[[10]]),annot(cars.egr[[10]]))
tlabels<-list(annot(trucks.no.tech[[10]]),annot(trucks.egr[[10]]))
hp.fp.ols<-	ggplot(data=dtt, aes(y=hpwt.crbwt.norm.CO2, x=Footprint..PU.Average, color=NHTSA.Defined.New.NHTSA.Car.Truck, size=Volume.2021, weight=Weight))+
	geom_point(aes(shape=1))+
	geom_smooth(method=lm, formula= y~ poly(x,1, raw=TRUE),se=FALSE)+
	labs(x="Footprint", y="WT/FP Normalized CO2", colour="C/T")+
	opts(title=paste(tit,"WT/FP Normalized CO2  v. Footprint (Wt. OLS)"))	+
	coord_cartesian(xlim = c(35,80), ylim=c(0,500))+
	annotate(geom="text",x=50,y=50, colour="black", label= paste("Car:",clabels[[hj]]) , size=3 ) +
	annotate(geom="text", x = 50, y= 100, colour="black", label=paste("Trucks:",tlabels[[hj]]) , size=3 ) 
	

	return(list(base.chart,base.ols,hp.fp.ols))
	
}



physics<-function(){
#Plot of WT/ v. FP - Weighted OLS
p2<-ggplot(data=fleet, aes(y=I(fleet$Curb.Weight), x=fleet$Footprint..PU.Average, size=Volume.2021,  weight=fleet$Weight, shape=1))+
	geom_point(aes(shape=1))+
	labs(x="Footprint", y="WT", colour="C/T", size="2021 Sales")+
	geom_smooth(method=lm, formula= y~ poly(x,1, raw=TRUE), se=FALSE)+
	geom_smooth(span=1,colour="red")+
	facet_wrap(~NHTSA.Defined.New.NHTSA.Car.Truck)+
	opts(title="WT v. FP - Weighted OLS and Loess Fit")

#Plot of WT/FP v. FP - Weighted OLS
p1<-ggplot(data=fleet, aes(y=I(fleet$Curb.Weight/fleet$Footprint..PU.Average), x=fleet$Footprint..PU.Average, size=Volume.2021,  weight=fleet$Weight))+
	geom_point(aes(shape=1))+
	labs(x="Footprint", y="WT/FP", colour="C/T", size="2021 Sales")+
	geom_smooth(method=lm, formula= y~ poly(x,1, raw=TRUE), se=FALSE)+
	geom_smooth(span=1,colour="red")+
	facet_wrap(~NHTSA.Defined.New.NHTSA.Car.Truck)+
	opts(title="WT/FP v. FP - Weighted OLS and Loess Fit")


	
# Plot of HP/WT v. FP - Weighted OLS
p4<-ggplot(data=fleet, aes(y=I(fleet$Horsepower/fleet$Curb.Weight), x=fleet$Footprint..PU.Average, size=Volume.2021, weight=fleet$Weight))+
	geom_point(aes(shape=1))+
	geom_smooth(method=lm, formula= y~ poly(x,1, raw=TRUE), se=FALSE)+
	facet_wrap(~NHTSA.Defined.New.NHTSA.Car.Truck)+
	labs(x="Footprint", y="HP/WT", colour="C/T", size="2021 Sales")+
	opts(title="HP/WT v. FP - Weighted OLS")



return(list(p1,p2,p4))
}

cars.no.tech<-analysis("C",1,"No Tech")
trucks.no.tech<-analysis("T",1,"No Tech")
cars.egr<-analysis("C",2,"EGR")
trucks.egr<-analysis("T",2,"EGR")

fleet3<-data.frame(rbind(cars.no.tech[[1]], trucks.no.tech[[1]]))
fleet4<-data.frame(rbind(cars.egr[[1]], trucks.egr[[1]]))

pltcars.no.tech<-mregres("C",1)
plttks.no.tech<-mregres("T",1)
pltcars.egr<-mregres("C",2)
plttks.egr<-mregres("T",2)

pdf(paste("WT.FP_curves_for_analysis-",Sys.Date(),".pdf",sep=""))
dchart(fleet4, "Max ICE Tech -",2)
physics()
fleet$sales.2008<-fleet$Total.Production.Volume

ggplot(data=fleet, aes(y=CO2, x=Footprint..PU.Average, size=sales.2008))+
	geom_point(shape=1)+
	labs(x="Footprint", y="2008 CO2", colour="Car/Truck", size="2008 Sales")+
	opts(title=paste("2008 CO2 v. Footprint"))+
	facet_wrap(~NHTSA.Defined.New.NHTSA.Car.Truck)+
	coord_cartesian(xlim = c(35,80), ylim=c(0,700))


summary(cars.egr[[7]])
summary(trucks.egr[[7]])
sd(cars.egr[[1]]$CO2)
sd(cars.egr[[1]]$CO2.Package.1)
sd(trucks.egr[[1]]$CO2)
sd(trucks.egr[[1]]$CO2.Package.1)


#Get Densities
man.sales<-aggregate((fleet$Volume.2021), by=list(fleet$Our.Class), sum)
colnames(man.sales)<- c("Our.Class", "Man.2021.Sales") 

man.footprint<-stack(by(fleet,list(fleet$Our.Class),function(subset)weighted.mean(subset$Footprint..PU.Average, subset$Volume.2021), simplify=TRUE))
colnames(man.footprint)<- c("Avg.Footprint","Our.Class") 

	tyu2<-merge(man.sales,fleet)	
	tyu2$man.proportion<-tyu2$Volume.2021/tyu2$Man.2021.Sales
	tyu2$wt.fp.proportion<-tyu2$Curb.Weight/tyu2$Footprint..PU.Average*tyu2$man.proportion
	manwtfp<-aggregate((tyu2$wt.fp.proportion), by=list(tyu2$Our.Class), sum)
	colnames(manwtfp) <- c("Our.Class", "Avg.Density")
	manwtfp<- transform(manwtfp, Our.Class=reorder(Our.Class,Avg.Density)) 
	testmerge<-merge(manwtfp,man.sales)
	manwtfp<-merge(testmerge, man.footprint)
	#wt/fp averages
	
ct<-signif(sum(cars.egr[[1]]$Curb.Weight/cars.egr[[1]]$Footprint..PU.Average*cars.egr[[1]]$Weight),digits=3)
dt<-signif(sum(trucks.egr[[1]]$Curb.Weight/trucks.egr[[1]]$Footprint..PU.Average*trucks.egr[[1]]$Weight), digits=3)

	ggplot(data=manwtfp,aes(x=Avg.Footprint, y=Avg.Density,label=Our.Class))+
		geom_point(size=1)+
		geom_text(hjust=-.05, angle=30, size=3)+
		labs(x="Footprint",y="WT/FP")+
		opts(title=paste("WT/FP by Vehicle Class"))+ 
		opts(axis.text.x=theme_text(hjust=1))+
		geom_hline(yintercept=ct, linetype=2, colour="red")+
		annotate(geom="text", x = 65, y= 71, colour="red",label=paste("Car fleet avg.:",ct), size=3)+
		geom_hline(yintercept=dt, linetype=2, colour="blue")+
		annotate(geom="text", x = 65, y= 83, colour="blue",label=paste("Truck fleet avg.:",dt), size=3)+
		scale_x_continuous(limits=c(40,70))+
		scale_y_continuous(limits=c(65,100))
	
#Footprint Histogram	
	ggplot(data=fleet,aes(x=Footprint..PU.Average, weight=Volume.2021))+
		geom_bar(color="darkblue", fill="white", binwidth=1)+
		facet_wrap(~NHTSA.Defined.New.NHTSA.Car.Truck,ncol=1)+
		labs(x="Footprint",y="MY 2021 Projected Sales")+
		opts(title=paste("Footprint Distribution by Car and Truck"))+
		geom_vline(xintercept=41, linetype=2, colour="darkgreen")+
		geom_vline(xintercept=56, linetype=2, colour="darkgreen")+
		geom_vline(xintercept=74, linetype=4, colour="red")+
		geom_vline(xintercept=41, linetype=4, colour="red")

#wt/fp co2
fleetwadj<-rbind(cars.egr[[1]],trucks.egr[[1]])

ggplot(data=fleetwadj, aes(y=hpwt.crbwt.norm.CO2/8887, x=Footprint..PU.Average, size=Volume.2021, weight=Weight))+
	geom_point(aes(shape=1))+
	facet_wrap(~NHTSA.Defined.New.NHTSA.Car.Truck)+
	geom_smooth(method=lm, formula= y~ poly(x,1, raw=TRUE), se=FALSE)+
	labs(x="Footprint", y="GPM", colour="C/T")+
	opts(title=paste("WT/FP Normalized Fuel Consumption  v. Footprint"))+
	scale_y_continuous(limits=c(0,0.06))	

dev.off()
