### Class examples from Week 2: Data
### Ken Benoit Jan 2010

## FIRST YOU NEED TO DOWNLOAD THE FILE dail2002.Rdata
## FROM http://www.kenbenoit.net/courses/iqrm/dail2002.Rdata
##
## let's say you download it to your Desktop
##
## then you need to do this (may only work for Mac OS)
setwd("~/Desktop")
## or change the working directory from a pull-down menu

load("http://www.kenbenoit.net/courses/iqrm/dail2002.Rdata")
class(dail2002)
attach(dail2002)

# frequency distribution of candidates by party
table(party)
# frequency distribution of candidates by party as proportion
prop.table(table(party))
# put them together in a single table using cbind()
cbind(table(party), prop.table(table(party))*100)

# two-way contingency table of party by incumbency status
table(party, incumbf)
# sum the columns using apply()
apply(table(party, incumbf), 2, sum)


# produces too many values!
table(spend_total)

# this cuts spending into 10 ranges
table(cut(spend_total, breaks=10, dig.lab=6))

# this cuts spending into ranges that we define
t1 <- table(cut(spend_total, breaks=c(0,5000,10000,20000,30000,40000),
                dig.lab=6))
t1

# this tabulates spending intervals by candidate status
t2 <- table(incumbf, cut(spend_total,
                         breaks=c(0,5000,10000,20000,30000,40000),
                         dig.lab=6))
t2
# we can "transpose" the table using t()
t(t2)

# pie chart of candidates by party
pie(table(party))
# pie chart of spending by party
tapply(spend_total, party, sum)
pie(tapply(spend_total, party, sum))
pie(tapply(spend_total, party, sum), main="Spending Share by Party")

# barplot of candidates by party
barplot(table(party))

# barplot of spending intervals from above example
barplot(t1)

# barplot of spending intervals by cand status from above
barplot(t2)
barplot(t2, beside=T)
barplot(t2, beside=T, legend=T)

# box plots of spending by incumbency, then by party
plot(incumbf, spend_total)
plot(party, spend_total)

# histogram of spending
hist(spend_total, xlab="Candidate Spending", main="")

# density plots to show skew
plot(density(spend_total))
plot(density(votes1st, na.rm=T))

# overlay histogram with density plot
hist(spend_total, xlab="Candidate Spending", main="", freq=F)
lines(density(spend_total))

# scatterplot!
plot(spend_total, votes1st)