#CA Developmentally Disabled Expenditures Analysis

#Read in data
CAExpend <- read.csv("C:/Users/CMalone/Desktop/DSCI Workshop All Content/CA_DDExpenditures/CA_DDExpenditures.csv")
View(CAExpend)

#Means only on Ethnicity -- age *ignored* here
output<-aggregate(CAExpend$Expenditures,by=list(CAExpend$Ethnicity),mean)
#Convert to vector for barplot() function
output.vector <- as.vector(output[,2])
names(output.vector) <- output[, 1]
barplot(output.vector[c(4,8)],ylim=c(0,60000))

#Get a subset and reconstruct plot 
CAExpend.Subset<-CAExpend[CAExpend$Ethnicity %in% c("Hispanic","White not Hispanic"),]
output<-aggregate(CAExpend.Subset$Expenditures,by=list(CAExpend.Subset$Ethnicity),mean)
output.vector <- as.vector(output[,2])
names(output.vector) <- output[, 1]
barplot(output.vector,ylim=c(0,60000))

#Means only on Ethnicity AND AgeCohort -- age considered here
output2<-aggregate(CAExpend.Subset$Expenditures,by=list(CAExpend.Subset$Ethnicity,CAExpend.Subset$AgeCohort),mean)
#Convert to vector for barplot() function
output2.vector <- as.vector(output2[,3])
#Make names pretty, use H for Hispanic, W for White not Hispanic, clean up spaces on age cohort
names(output2.vector) <- paste(substr(output2[,1],1,1),gsub(" ","",output2[,2]),sep=":")
barplot(output2.vector,ylim=c(0,60000),las=2)

######################
#Change order of Age on x-axis

#Correct number of factor levels for ethnicity; not necessary for plotting but data quality issue
CAExpend.Subset$Ethnicity<-factor(CAExpend.Subset$Ethnicity,levels=c("Hispanic","White not Hispanic"))

#Redefine order of Age Cohort, necessary to get order correct for Age
CAExpend.Subset$AgeCohort<-factor(CAExpend.Subset$AgeCohort,levels=c(" 0 - 5", "6-12", "13-17", "18-21", "22-50"," 51 +"))

#Obtain average expenditures using new ordering of AgeCohort
output3<-aggregate(CAExpend.Subset$Expenditures,by=list(CAExpend.Subset$Ethnicity,CAExpend.Subset$AgeCohort),mean)
#Convert to vector for barplot() function
output3.vector <- as.vector(output3[,3])

#Make names pretty, use H for Hispanic, W for White not Hispanic, clean up spaces on age cohort
names(output3.vector) <- paste(substr(output3[,1],1,1),gsub(" ","",output3[,2]),sep=":")
barplot(output3.vector,ylim=c(0,60000),las=2)

#Adding vertical lines, save divisions into barplot.output
barplot.output<-barplot(output3.vector,ylim=c(0,60000),las=2)
#Add title, 75% of default font size
title(main="Comparing Hispanics to White - Age Adjusted", cex.main=0.75)

#Use abline() and v= for vertical lines
abline(v=c(mean(barplot.output[2:3,1]),mean(barplot.output[4:5,1]),mean(barplot.output[6:7,1]),mean(barplot.output[8:9,1]),mean(barplot.output[10:11,1])),col="gray")