CANSIM2.Rnw

\documentclass{article}
\usepackage[margin=0.01in,paperwidth=13in]{geometry}
%Use geometry package. Make page margins as small as possible to put as much stuff on the page as possible
%Make the paper width wider than normal in order to fit everything on the page
\begin{document}

<<echo=FALSE,results='asis',message=FALSE,warning=FALSE>>=
library(xtable)
library(ggplot2)
library(reshape)
library(stringr)
library(scales)
library(Hmisc)
library(grid)
fileUrl <- "http://www20.statcan.gc.ca/tables-tableaux/cansim/csv/02020802-eng.zip"
temp <- tempfile()
download.file(fileUrl, temp)
data.test <- read.csv(unz(temp, "02020802-eng.csv"))
unlink(temp)
###Above I have grabbed the data directly from the website
data.test<-data.test[,c(-7,-6)] #Dropping "vector" columns
colnames(data.test)<-c("Year","Geography","Line","Statistic","Population","Value") #labelling

geo<-c("Atlantic provinces","Newfoundland and Labrador", #Creating vector for geographies of interest
       "New Brunswick","Nova Scotia","Prince Edward Island",
       "Prairie provinces","Alberta","Manitoba","Saskatchewan",
       "British Columbia","Ontario","Quebec","Canada")
population<-c("Females","Males","Persons under 18 years", #Creating vector for populations of interest
              "Persons 18 to 64 years","Persons 65 years and over",
              "Persons in economic families",
              "Persons under 18 years in female lone-parent families",
              "Persons under 18 years in two-parent families",
              "Unattached individuals","All persons")
data2<-subset(data.test,!is.na(data.test$Population)&!is.na(data.test$Geography)& #Drop NA values
                (data.test$Geography %in% geo)) #Keep all rows where Geography column matches
  
#Small data cleaning...reordering and renaming certain factor levels
data2$Value<-as.numeric(as.character(data2$Value)) #Change value from factor to numeric
#Below I am reordering the factor levels of Geography and Population since I can't do it within Hmisc functions
data2$Geography<-ordered(data2$Geography,levels=c("Atlantic provinces","Newfoundland and Labrador",
                                                  "New Brunswick","Nova Scotia","Prince Edward Island",
                                                  "Prairie provinces","Alberta","Manitoba","Saskatchewan",
                                                  "British Columbia","Ontario","Quebec","Canada"))
data2$Population<-ordered(data2$Population,levels=c("Females","Males","Persons under 18 years",
                                                    "Persons 18 to 64 years","Persons 65 years and over",
                                                    "Persons in economic families","Persons under 18 years in female lone-parent families","Persons under 18 years in two-parent families","Unattached individuals","All persons",
              "Females (x 1,000)","Males (x 1,000)","Persons under 18 years (x 1,000)",
              "Persons 18 to 64 years (x 1,000)","Persons 65 years and over (x 1,000)",
              "Persons in economic families (x 1,000)",
              "Persons under 18 years in female lone-parent families (x 1,000)",
              "Persons under 18 years in two-parent families (x 1,000)",
              "Unattached individuals (x 1,000)","All persons (x 1,000)")
                                                    )
data2$Population<-revalue(data2$Population,c("Persons under 18 years in female lone-parent families"="Child in single mother families","Persons under 18 years in two-parent families"="Child in two-parent families"))
#End of data cleaning

#1a. Pov. rates across time and Population 2000 onwards for LICO
#We subset the data to specify Overall canada geography, ratios rather than counts and the proper Line
#I am asking Hmisc to summarize Value (Pov rates) by Population and by Year.
#Method cross is the only way to map 3 variables at once. It automatically resorts to aggregation via MEAN
#However given that we already have individual Pov rates for all Population-Year, Hmisc does not aggregate
#anything. Essentially, this code is here to be readily applied to more difficult datasets, but it is not
#the most direct approach for the CANSIM data
#I cannot remove the counts (N) from the table.
tab1a<-with(subset(data2,Year>=2000&Geography=="Canada"&Statistic=="Percentage of persons in low income"&(Population %in% population)&
              as.character(data2$Line)=="Low income cut-offs after tax, 1992 base"),
            summary(Value~Population+factor(Year),method='cross',overall=FALSE,prn=FALSE,prnmiss=FALSE))
latex(tab1a,file="",caption="LICO Poverty rates by Population",caption.loc="bottom")
#1b. Pov. rates across time and Population 2000 onwards for LIM
tab1b<-with(subset(data2,data2$Year>=2000&data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&(Population %in% population)&
              data2$Line=="Low income measure after tax"),
     summary(Value~Population+factor(Year),method='cross',overall=FALSE))
latex(tab1b,file="",caption="LIM Poverty rates by Population",caption.loc="bottom")
#1c. Pov. rates across time and Population 2000 onwards for MBM
tab1c<-with(subset(data2,data2$Year>=2000&data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&(Population %in% population)&
              data2$Line=="Market basket measure, 2011 base"),
     summary(Value~Population+factor(Year),method='cross',overall=FALSE))
latex(tab1c,file="",caption="MBM Poverty rates by Population",caption.loc="bottom")
#2a. Pov rates across time and Geography 2000 onwards for LICO
#We subset the data to specify overall All persons population, ratios rather than counts and the proper line
#I ask Hmisc to evaluate Value (pov rates) by Geography and Year
tab2a<-with(subset(data2,data2$Year>=2000&data2$Population=="All persons"&data2$Statistic=="Percentage of persons in low income"&
              data2$Line=="Low income cut-offs after tax, 1992 base"),
     summary(Value~Geography+factor(Year),method='cross',overall=FALSE,prn=FALSE,prnmiss=FALSE))
latex(tab2a,file="",caption="LICO Poverty by Province",caption.loc="bottom")
#2b. Pov rates across time and Geography 2000 onwards for LIM
tab2b<-with(subset(data2,data2$Year>=2000&data2$Population=="All persons"&data2$Statistic=="Percentage of persons in low income"&
              data2$Line=="Low income measure after tax"),
     summary(Value~Geography+factor(Year),method='cross',overall=FALSE))
latex(tab2b,file="",caption="LIM Poverty by Province",caption.loc="bottom")
#2c. Pov rates across time and Geography 2000 onwards for MBM
tab2c<-with(subset(data2,data2$Year>=2000&data2$Population=="All persons"&data2$Statistic=="Percentage of persons in low income"&
              data2$Line=="Market basket measure, 2011 base"),
     summary(Value~Geography+factor(Year),method='cross',overall=FALSE))
latex(tab2c,file="",caption="MBM Poverty by Province",caption.loc="bottom")
#3. Pov rates across time and Line
#we subset the data to specify overall ALL persons population, ratios rather than counts and Overall canada geography
#Hmisc is being difficult, I can't edit individual elements within Hmisc function, so I have to do it out here
###Little code to make my life easier
data2$Lines<-ordered(data2$Line,levels=c("Low income cut-offs after tax, 1992 base",
                                        "Low income measure after tax",
                                        "Market basket measure, 2011 base",
                                        "Low income cut-offs before tax, 1992 base"),
                    labels=c("LICO","LIM","MBM","LICO before tax"))
###But Not necessary
#Now I ask Hmisc to evaluate Value (pov rates) by the Poverty measuring lines and Years
tab3a<-with(subset(data2,data2$Year>=2000&data2$Population=="All persons"&data2$Statistic=="Percentage of persons in low income"&
                    data2$Geography=="Canada"),
           summary(Value~Lines+factor(Year),method='cross',prn=FALSE,prnmiss=FALSE,overall=FALSE),rowlabel="Line")
latex(tab3a,file="",caption="Poverty by Measuring Line",caption.loc="bottom")
@
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#NOTICE
#LICO before tax doesn't seem to be comparable to other measures, and so I have found no use for it.
#I will exclude it from the data. If you require LICO before tax, omit the following line of code
data2<-subset(data2,!data2$Line=="Low income cut-offs before tax, 1992 base")


#1. What is the overall canadian poverty trend over time?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     data2$Population=="All persons"),
       aes(x=Year,y=Value/100,colour=Line,group=Line))+ #Year on x-axis, pov rate/100 on y-axis and each line gets a different colour
  ylab("Poverty Rate")+ggtitle("Canadian Poverty Measures")+
  geom_line()+ #All elements of geom_line taken from ggplot aesthetics arguments
  geom_point()+ #All elements of geom_point taken from ggplot aesthetics arguemnts
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.01),labels=percent)+
  #Give a tick to every 1% pov rate on the y axis and convert pov rates (0<x<1) into percentages
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  #Legend reads left to right on top of graph and angle x-axis ticks to 45 degrees
  scale_x_continuous(breaks=seq(min(data2$Year,na.rm=TRUE),max(data2$Year,na.rm=TRUE),2))
  #Give a tick to every 2 years on the x-axis, starting from the minimum Year to the maximum Year
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#2. How does the overall trend differ with the child poverty trend?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     (data2$Population=="All persons"|data2$Population=="Persons under 18 years")),
       aes(x=Year,y=Value/100,colour=Line,linetype=Population))+
  geom_line()+
  ylab("Poverty Rate")+ggtitle("Canadian Child Poverty Measures")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.01),labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=seq(min(data2$Year,na.rm=TRUE),max(data2$Year,na.rm=TRUE),2))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#3a. What are the age-based poverty rates?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     !data2$Line=="Market basket measure, 2011 base"&
                     (data2$Population=="Persons under 18 years"|
                        data2$Population=="Persons 18 to 64 years"|data2$Population=="Persons 65 years and over")),
       aes(x=Year,y=Value/100,colour=Population))+
  geom_line()+ # All elements from geom_line taken from ggplot aesthetics arguments
  facet_grid(.~Line)+ #Split the plot into side-by-side plots using different Poverty measuring lines
  ylab("Poverty Rate")+xlab("Year")+ggtitle("Poverty across ages")+ #Naming
  #Give a tick to every 2% pov rate on the y axis from 0 to the maximum Poverty value and change values to percents
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.02),labels=percent)+
  #Put legend on top and read from left to right
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#3b. How do age-based poverty rates compare?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     !data2$Line=="Market basket measure, 2011 base"&
                     (data2$Population=="Persons under 18 years"|data2$Population=="Persons 18 to 64 years"|
                        data2$Population=="Persons 65 years and over")),
       aes(x=Year,y=Value/100,fill=Population))+geom_area(stat="identity")+facet_grid(.~Line)+
  ylab("Poverty Rate")+ggtitle("Comparing Poverty across ages")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.1),labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
    scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#3c. What is the poverty variance across ages?
ggplot(data=subset(data2,!data2$Line=="Market basket measure, 2011 base"&data2$Geography=="Canada"&
                     data2$Statistic=="Percentage of persons in low income"&
                     (data2$Population=="Persons under 18 years"|
                        data2$Population=="Persons 18 to 64 years"|data2$Population=="Persons 65 years and over")),
       aes(x=Population,y=Value/100,fill=Population))+ #Put different Populations on x axis and Pov rate on y axis and create/fill additional boxplots for every different Population
  geom_boxplot()+
  facet_grid(.~Line)+ #Split plot into side-by-side plots using different Povert measuring lines
  ggtitle("Poverty variance across ages")+ylab("Poverty Rate")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.05),labels=percent)+
  theme(legend.direction="horizontal",legend.position="top")+
  scale_x_discrete(labels=c("Children","Adults","Elderly"))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#3d. What is the age of the poor?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Number of persons in low income"&
                   !data2$Line=="Market basket measure, 2011 base"&
                     (data2$Population=="Persons under 18 years (x 1,000)"|
                        data2$Population=="Persons 18 to 64 years (x 1,000)"|
                        data2$Population=="Persons 65 years and over (x 1,000)")),
       aes(x=Year,y=Value,fill=Population))+geom_area(stat="identity",position="fill")+
         facet_grid(.~Line)+ylab("Composition of the poor")+ggtitle("Age of the Poor")+
  scale_y_continuous(labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_fill_discrete(labels=c("Persons under 18 years","Persons 18 to 64 years",
                               "Persons 65 years and over"))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))

@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#4a. What are the gender based poverty rates?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     !data2$Line=="Market basket measure, 2011 base"&
                     (data2$Population=="Males"|data2$Population=="Females")),
       aes(x=Year,y=Value/100,colour=Population))+geom_line()+facet_grid(.~Line)+
  ylab("Poverty rate")+ggtitle("Poverty across genders")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.01),labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#4b. How do gender-based poverty rates compare?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     !data2$Line=="Market basket measure, 2011 base"&
                     (data2$Population=="Males"|data2$Population=="Females")),
       aes(x=Year,y=Value/100,fill=Population))+geom_area(stat="identity")+facet_grid(.~Line)+
  ylab("Poverty rate")+ggtitle("Comparing Poverty across genders")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.1),labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#4c. What is the poverty variance across genders?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     !data2$Line=="Market basket measure, 2011 base"&
                     (data2$Population=="Males"|data2$Population=="Females")),
       aes(x=Population,y=Value/100,fill=Population))+geom_boxplot()+facet_grid(.~Line)+
  ylab("Poverty rate")+ggtitle("Poverty variance across genders")+ guides(fill=FALSE)+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.01),labels=percent)
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#4d. What is the gender of the poor?
ggplot(data=subset(data2,data2$Geography=="Canada"&data2$Statistic=="Number of persons in low income"&
                     !data2$Line=="Market basket measure, 2011 base"&
                     (data2$Population=="Males (x 1,000)"|data2$Population=="Females (x 1,000)")),
       aes(x=Year,y=Value,fill=Population))+geom_area(stat="identity",position="fill")+
  facet_grid(.~Line)+ylab("Composition of the poor")+ggtitle("Gender of the Poor")+
  scale_y_continuous(labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\clearpage
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#6a. What are the broader geographical based poverty rates?
ggplot(data=subset(data2,(data2$Geography=="Atlantic provinces"|
                            data2$Geography=="Prairie provinces"|data2$Geography=="Ontario"|
                            data2$Geography=="British Columbia"|data2$Geography=="Quebec")&
                     !data2$Line=="Market basket measure, 2011 base"&data2$Population=="All persons"&
                     data2$Statistic=="Percentage of persons in low income"),
       aes(x=Year,y=Value/100,colour=Geography))+geom_line()+facet_grid(.~Line)+
  ylab("Poverty Rate")+ggtitle("Poverty aross broader geography")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.02),labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#6b. How do broader geographical-based poverty rates compare?
ggplot(data=subset(data2,(data2$Geography=="Atlantic provinces"|data2$Geography=="Prairie provinces"|
                            data2$Geography=="Ontario"|data2$Geography=="British Columbia"|
                            data2$Geography=="Quebec")&
                     !data2$Line=="Market basket measure, 2011 base"&data2$Population=="All persons"&
                     data2$Statistic=="Percentage of persons in low income"),
       aes(x=Year,y=Value/100,fill=Geography))+geom_area(stat="identity")+facet_grid(.~Line)+
  ylab("Poverty Rate")+ggtitle("Comparing Poverty aross broader geography")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.1),labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#6c. What is the poverty variance across broader geography?
ggplot(data=subset(data2,!data2$Line=="Market basket measure, 2011 base"&data2$Population=="All persons"&
                     (data2$Geography=="Canada"|data2$Geography=="Atlantic provinces"|
                        data2$Geography=="Prairie provinces"|data2$Geography=="Ontario"|
                        data2$Geography=="British Columbia"|data2$Geography=="Quebec")&
                     data2$Statistic=="Percentage of persons in low income"),
       aes(x=Geography,y=Value/100,fill=Geography))+geom_boxplot()+facet_grid(.~Line)+
  ggtitle("Poverty variance across broader geography")+ylab("Poverty Rate")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),0.05),labels=percent)+
  theme(legend.direction="horizontal",legend.position="top")+
  scale_x_discrete(breaks=NULL)
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#6d. What is the broader geography of the poor?
ggplot(data=subset(data2,(data2$Geography=="Atlantic provinces"|data2$Geography=="Prairie provinces"|
                            data2$Geography=="Ontario"|data2$Geography=="British Columbia"|
                            data2$Geography=="Quebec")&
                     !data2$Line=="Market basket measure, 2011 base"&
                     data2$Population=="All persons (x 1,000)"&
                     data2$Statistic=="Number of persons in low income"),
       aes(x=Year,y=Value,fill=Geography))+geom_area(stat="identity",position="fill")+
  facet_grid(.~Line)+ylab("Composition of the poor")+ggtitle("Broader Geography of the Poor")+
  scale_y_continuous(labels=percent)+
  theme(legend.direction="horizontal",legend.position="top",axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#5a. What are the geographical based poverty rates?
ggplot(data=subset(data2,(!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&
                     !data2$Geography=="Prairie provinces")&data2$Statistic=="Percentage of persons in low income"&
                     data2$Population=="All persons"&!data2$Line=="Market basket measure, 2011 base"),
       aes(x=Year,y=Value/100,colour=Geography))+geom_line()+facet_grid(Line~.)+
  ylab("Poverty Rate")+ggtitle("Poverty aross provinces")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.05),labels=percent)+
  theme(axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#5b. How do geographical-based poverty rates compare?
ggplot(data=subset(data2,(!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&
                            !data2$Geography=="Prairie provinces")&data2$Statistic=="Percentage of persons in low income"&
                     data2$Population=="All persons"&!data2$Line=="Market basket measure, 2011 base"),
       aes(x=Year,y=Value/100,fill=Geography))+geom_area(stat="identity")+facet_grid(.~Line)+
  ylab("Poverty Rate")+ggtitle("Comparing Poverty aross provinces")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),.1),labels=percent)+
  theme(axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#5c. What is the poverty variance across geography?
ggplot(data=subset(data2,!data2$Line=="Market basket measure, 2011 base"&data2$Population=="All persons"&
                     (!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&
                        !data2$Geography=="Prairie provinces")&data2$Statistic=="Percentage of persons in low income"),
       aes(x=Geography,y=Value/100,fill=Geography))+geom_boxplot()+facet_grid(Line~.)+
  ggtitle("Poverty variance across provinces")+ylab("Poverty Rate")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),0.05),labels=percent)+
  scale_x_discrete(labels=c("NL","NB","NS","PI","AB","MB","SK","BC","ON",'QC'))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#5d. What is the geography of the poor?
ggplot(data=subset(data2,(!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&
                            !data2$Geography=="Prairie provinces")&data2$Statistic=="Number of persons in low income"&
                     data2$Population=="All persons (x 1,000)"&
                     !data2$Line=="Market basket measure, 2011 base"),
       aes(x=Year,y=Value,fill=Geography))+geom_area(stat="identity",position="fill")+facet_grid(.~Line)+
  ylab("Composition of the poor")+ggtitle("Geography of the poor")+scale_y_continuous(labels=percent)+
  theme(axis.text.x=element_text(angle=45))+
  scale_x_continuous(breaks=c(min(data2$Year,na.rm=TRUE),seq(1980,max(data2$Year,na.rm=TRUE),5)))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#2011

#7a. Which age group currently has the highest  poverty rate [PARETO PLOT]
#LIM
ggplot(data=subset(data2,data2$Year==max(data2$Year)&data2$Line=="Low income measure after tax"
                   &data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     (data2$Population=="Persons under 18 years"|data2$Population=="Persons 18 to 64 years"|
                        data2$Population=="Persons 65 years and over")),
       aes(x=reorder(Population,-Value),y=Value/100,fill=Population))+geom_bar(stat="identity")+
  ylab("Poverty Rate")+xlab("Population")+ggtitle("2011 LIM Poverty rate-age Pareto plot")+guides(fill=FALSE)+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),0.01),labels=percent)
#Pareto plots rely on descending order of bars therefore I can't facet between LIM-MBM and LICO since their orders are different
#Now LICO
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
ggplot(data=subset(data2,data2$Year==max(data2$Year)&data2$Line=="Low income cut-offs after tax, 1992 base"&
                     data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     (data2$Population=="Persons under 18 years"|data2$Population=="Persons 18 to 64 years"|
                        data2$Population=="Persons 65 years and over")),
       aes(x=reorder(Population,-Value),y=Value/100,fill=Population))+geom_bar(stat="identity")+
  ylab("Poverty Rate")+xlab("Population")+ggtitle("2011 LICO Poverty rate-age Pareto plot")+guides(fill=FALSE)+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),0.01),labels=percent)
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#Now MBM
ggplot(data=subset(data2,data2$Year==max(data2$Year)&data2$Line=="Market basket measure, 2011 base"&
                     data2$Geography=="Canada"&data2$Statistic=="Percentage of persons in low income"&
                     (data2$Population=="Persons under 18 years"|data2$Population=="Persons 18 to 64 years"|
                        data2$Population=="Persons 65 years and over")),
       aes(x=reorder(Population,-Value),y=Value/100,fill=Population))+geom_bar(stat="identity")+
  ylab("Poverty Rate")+xlab("Population")+ggtitle("2011 MBM Poverty rate-age Pareto plot")+guides(fill=FALSE)+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),0.01),labels=percent)
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#7b. Which age group currently has the highest poverty population? [FULL PARETO PLOT]
a<-arrange(subset(data2,data2$Year==max(data2$Year)&data2$Geography=="Canada"&
            data2$Statistic=="Number of persons in low income"&
                    (data2$Population=="Persons under 18 years (x 1,000)"|
                       data2$Population=="Persons 18 to 64 years (x 1,000)"|
               data2$Population=="Persons 65 years and over (x 1,000)")&
           data2$Line=="Low income cut-offs after tax, 1992 base"),desc(Value)) 
a$cumsum<-cumsum(a$Value)
#First lets arrange our subset of data (since we want to examine age group, we leave all values for age group,
#and exclude or control for others). Select the Line to use. Taking the subset, arrange it by Value in descending order.
#Then create a culumulative sum variable in that order
#Now on the x-axis, plot in Age groups in descending order of value
#LICO
g1<-ggplot(data=a,
       aes(x=reorder(Population,-Value)))+ #Arrange Population on x-axis from highest to lowest
       geom_bar(stat="identity",aes(y=Value,fill=Population))+ #Measured by pov rate, each Population gets a bar
  geom_line(aes(y=cumsum,group=1))+#Plot the cumsum line
  geom_point(aes(y=cumsum))+ #Plot the points of the cumsum line
  ylab("Total Poverty Count (per 1,000 persons)")+xlab("Population")+ggtitle("2011 LICO")+
  guides(fill=FALSE)+scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),500))+
  scale_x_discrete(labels=c("Adults","Children","Elderly"))
#LIM
b<-arrange(subset(data2,data2$Year==max(data2$Year)&data2$Geography=="Canada"&
                    data2$Statistic=="Number of persons in low income"&
                    (data2$Population=="Persons under 18 years (x 1,000)"|
                       data2$Population=="Persons 18 to 64 years (x 1,000)"|
                       data2$Population=="Persons 65 years and over (x 1,000)")&
                    data2$Line=="Low income measure after tax"),desc(Value)) 
b$cumsum<-cumsum(b$Value)
g2<-ggplot(data=b,
       aes(x=reorder(Population,-Value)))+
  geom_bar(stat="identity",aes(y=Value,fill=Population))+
  geom_line(aes(y=cumsum,group=1))+geom_point(aes(y=cumsum))+
  ylab("Total Poverty Count (per 1,000 persons)")+xlab("Population")+ggtitle("2011 LIM")+
  guides(fill=FALSE)+scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),500))+
  scale_x_discrete(labels=c("Adults","Children","Elderly"))
#MBM
c<-arrange(subset(data2,data2$Year==max(data2$Year)&data2$Geography=="Canada"&
                    data2$Statistic=="Number of persons in low income"&
                    (data2$Population=="Persons under 18 years (x 1,000)"|
                       data2$Population=="Persons 18 to 64 years (x 1,000)"|
                       data2$Population=="Persons 65 years and over (x 1,000)")&
                    data2$Line=="Market basket measure, 2011 base"),desc(Value)) 
c$cumsum<-cumsum(c$Value)
g3<-ggplot(data=c,
       aes(x=reorder(Population,-Value)))+
  geom_bar(stat="identity",aes(y=Value,fill=Population))+
  geom_line(aes(y=cumsum,group=1))+geom_point(aes(y=cumsum))+
  ylab("Total Poverty Count (per 1,000 persons)")+xlab("Population")+ggtitle("2011 MBM")+
  guides(fill=FALSE)+scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),500))+
  scale_x_discrete(labels=c("Adults","Children","Elderly"))

#Creating custom function to place entirely seperate ggplot graphics together using "Grid" package
layout<-grid.layout(nrow=1,ncol=3)
vplayout<-function(...){
  grid.newpage()
  pushViewport(viewport(layout=layout))
}
subplot<-function(x,y) viewport(layout.pos.row=x,layout.pos.col=y)
mmplot<-function(a,b,c){
  vplayout()
  print(a,vp=subplot(1,1))
  print(b,vp=subplot(1,2))
  print(c,vp=subplot(1,3))
}
mmplot(g1,g2,g3)
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#8a. Which province currently has the highest poverty count? [PARETO]
#LICO
ggplot(data=subset(data2,data2$Year==max(data2$Year)&data2$Line=="Low income cut-offs after tax, 1992 base"&
                    (!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&!data2$Geography=="Prairie provinces")
                  &data2$Statistic=="Percentage of persons in low income"&data2$Population=="All persons"),
      aes(x=reorder(Geography,-Value),y=Value/100,fill=Geography))+
  geom_bar(stat="identity")+
  ylab("Poverty Rate")+xlab("Province")+ggtitle("2011 LICO Poverty-Province Pareto plot")+guides(fill=FALSE)+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),0.01),labels=percent)+
  scale_x_discrete(labels=c("BC","QC","ON","MN","NS","AB","NB","NL","SK","PI"))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#LIM
ggplot(data=subset(data2,data2$Year==max(data2$Year)&data2$Line=="Low income measure after tax"&
                     (!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&!data2$Geography=="Prairie provinces")
                   &data2$Statistic=="Percentage of persons in low income"&data2$Population=="All persons"),
       aes(x=reorder(Geography,-Value),y=Value/100,fill=Geography))+geom_bar(stat="identity")+
  ylab("Poverty Rate")+xlab("Province")+ggtitle("2011 LIM Poverty-Province Pareto plot")+guides(fill=FALSE)+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),0.01),labels=percent)+
  scale_x_discrete(labels=c("BC","PI","MN","QC","NL","NS","NB","ON","SK","AB"))

@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#MBM
ggplot(data=subset(data2,data2$Year==max(data2$Year)&data2$Line=="Market basket measure, 2011 base"&
                     (!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&!data2$Geography=="Prairie provinces")
                   &data2$Statistic=="Percentage of persons in low income"&data2$Population=="All persons"),
       aes(x=reorder(Geography,-Value),y=Value/100,fill=Geography))+geom_bar(stat="identity")+
  ylab("Poverty Rate")+xlab("Province")+ggtitle("2011 MBM Poverty-Province Pareto plot")+guides(fill=FALSE)+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),0.01),labels=percent)+
  scale_x_discrete(labels=c("BC","NS",'PI','NB','ON','NL','MN','QC','SK','AB'))
@
\end{center}
\end{figure}
\begin{figure}[ht]
\begin{center}
<<echo=FALSE,fig.width=12>>=
#8b. Which Province currently has the highest poverty population? [FULL PARETO PLOT]
b<-subset(data2,data2$Year==max(data2$Year)&(!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&
                                               !data2$Geography=="Prairie provinces")&
            data2$Statistic=="Number of persons in low income"&data2$Population=="All persons")
          #First create our subset data b which now contains all provinces but one age group "All persons"
#LICO
d<-arrange(subset(data2,data2$Year==max(data2$Year)&(!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&
                                                       !data2$Geography=="Prairie provinces")&
                    data2$Statistic=="Number of persons in low income"&
                    (data2$Population=="All persons (x 1,000)")&
                    data2$Line=="Low income cut-offs after tax, 1992 base"),desc(Value)) 
d$cumsum<-cumsum(d$Value)
#Now our subset consists of all provinces, but keeps Population at "All persons".
#Select the Line to use. Taking the subset, arrange it by Value in descending order.
#Then create a culumulative sum variable in that order
#Now on the x-axis, plot in Age groups in descending order of value
#LICO
g4<-ggplot(data=d,
       aes(x=reorder(Geography,-Value)))+
  geom_bar(stat="identity",aes(y=Value,fill=Geography))+
  geom_line(aes(y=cumsum,group=1))+geom_point(aes(y=cumsum))+
  ylab("Total Poverty Count (per 1,000 persons)")+xlab("Population")+ggtitle("2011 LICO")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),500))+
  scale_x_discrete(labels=c('ON','QC','BC','AB','MN','NS','SK','NB','NL','PI'))+guides(fill=F)

e<-arrange(subset(data2,data2$Year==max(data2$Year)&(!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&
                                                       !data2$Geography=="Prairie provinces")&
                    data2$Statistic=="Number of persons in low income"&
                    (data2$Population=="All persons (x 1,000)")&
                    data2$Line=="Low income measure after tax"),desc(Value)) 
e$cumsum<-cumsum(e$Value)
#Now our subset consists of all provinces, but keeps Population at "All persons".
#Select the Line to use. Taking the subset, arrange it by Value in descending order.
#Then create a culumulative sum variable in that order
#Now on the x-axis, plot in Age groups in descending order of value
g5<-ggplot(data=e,
       aes(x=reorder(Geography,-Value)))+
  geom_bar(stat="identity",aes(y=Value,fill=Geography))+
  geom_line(aes(y=cumsum,group=1))+geom_point(aes(y=cumsum))+
  ylab("Total Poverty Count (per 1,000 persons)")+xlab("Population")+ggtitle("2011 LIM")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),500))+
  scale_x_discrete(labels=c("ON",'QC','BC','AB','MN','NS','SK','NB','NL','PI'))+guides(fill=F)

f<-arrange(subset(data2,data2$Year==max(data2$Year)&(!data2$Geography=="Canada"&!data2$Geography=="Atlantic provinces"&
                                                       !data2$Geography=="Prairie provinces")&
                    data2$Statistic=="Number of persons in low income"&
                    (data2$Population=="All persons (x 1,000)")&
                    data2$Line=="Market basket measure, 2011 base"),desc(Value)) 
f$cumsum<-cumsum(f$Value)
#Now our subset consists of all provinces, but keeps Population at "All persons".
#Select the Line to use. Taking the subset, arrange it by Value in descending order.
#Then create a culumulative sum variable in that order
#Now on the x-axis, plot in Age groups in descending order of value
g6<-ggplot(data=f,
       aes(x=reorder(Geography,-Value)))+
  geom_bar(stat="identity",aes(y=Value,fill=Geography))+
  geom_line(aes(y=cumsum,group=1))+geom_point(aes(y=cumsum))+
  ylab("Total Poverty Count (per 1,000 persons)")+xlab("Population")+ggtitle("2011 MBM")+
  scale_y_continuous(breaks=seq(0,max(data2$Value,na.rm=TRUE),500))+
  scale_x_discrete(labels=c('ON','QC','BC','AB','MN','NS','SK','NB','NL','PI'))+guides(fill=F)

mmplot(g4,g5,g6)
@
\end{center}
\end{figure}
\end{document}