Data Management
- Original data came from intenet version of Sejong silok, summarized by Oh, Ki-Soo.
sejong.poll<-read.table("sejong_poll.txt",header=TRUE,sep="")
str(sejong.poll)
## 'data.frame': 44 obs. of 4 variables:
## $ counts: int 21 194 259 393 443 117 1123 71 29 5 ...
## $ vote : chr "yes" "no" "yes" "no" ...
## $ class : chr "high" "high" "third.current" "third.current" ...
## $ region: chr "Seoul" "Seoul" "Seoul" "Seoul" ...
sejong.poll
## counts vote class region
## 1 21 yes high Seoul
## 2 194 no high Seoul
## 3 259 yes third.current Seoul
## 4 393 no third.current Seoul
## 5 443 yes third.ex Seoul
## 6 117 no third.ex Seoul
## 7 1123 yes ordinary yuhu
## 8 71 no ordinary yuhu
## 9 29 yes chief gyunggi
## 10 5 no chief gyunggi
## 11 17076 yes ordinary gyunggi
## 12 236 no ordinary gyunggi
## 13 1 no high pyungan
## 14 6 yes chief pyungan
## 15 35 no chief pyungan
## 16 1326 yes ordinary pyungan
## 17 28474 no ordinary pyungan
## 18 17 yes chief hwanghae
## 19 17 no chief hwanghae
## 20 4454 yes ordinary hwanghae
## 21 15601 no ordinary hwanghae
## 22 2 no high chungcheong
## 23 35 yes chief chungcheong
## 24 26 no chief chungcheong
## 25 6982 yes ordinary chungcheong
## 26 14013 no ordinary chungcheong
## 27 5 yes chief kangwon
## 28 10 no chief kangwon
## 29 939 yes ordinary kangwon
## 30 6888 no ordinary kangwon
## 31 1 no high hamgil
## 32 3 yes chief hamgil
## 33 14 no chief hamgil
## 34 75 yes ordinary hamgil
## 35 7387 no ordinary hamgil
## 36 55 yes chief gyungsang
## 37 16 no chief gyungsang
## 38 36262 yes ordinary gyungsang
## 39 377 no ordinary gyungsang
## 40 2 no high jeolla
## 41 42 yes chief jeolla
## 42 12 no chief jeolla
## 43 29505 yes ordinary jeolla
## 44 257 no ordinary jeolla
- We need vote, class, region as factors. If you leave them as chr, it will be coerced to factor when you tabulate it according to alphabetical order, which is not what you want. So, use factor() to convert them. First, make a working copy vesion of sejong.poll
sejong.poll.2<-sejong.poll
sejong.poll.2$vote<-factor(sejong.poll.2$vote, levels=c("yes","no"), labels=c("yes","no"))
str(sejong.poll.2)
## 'data.frame': 44 obs. of 4 variables:
## $ counts: int 21 194 259 393 443 117 1123 71 29 5 ...
## $ vote : Factor w/ 2 levels "yes","no": 1 2 1 2 1 2 1 2 1 2 ...
## $ class : chr "high" "high" "third.current" "third.current" ...
## $ region: chr "Seoul" "Seoul" "Seoul" "Seoul" ...
- You can check that “labels=” is not necessary if same as levels. Continue with class and region.
sejong.poll.2$class<-factor(sejong.poll.2$class, levels=c("high","third.current", "third.ex", "chief", "ordinary"), labels=c("High","3rd.current", "3rd.former", "Chief", "Commons"))
sejong.poll.2$region<-factor(sejong.poll.2$region, levels=c("Seoul","yuhu", "gyunggi", "pyungan", "hwanghae", "chungcheong", "kangwon", "hamgil", "gyungsang", "jeolla"), labels=c("Seoul","Yuhu", "Gyunggi", "Pyungan", "Hwanghae", "Chungcheong", "Kangwon", "Hamgil", "Gyungsang", "Jeolla"))
str(sejong.poll.2)
## 'data.frame': 44 obs. of 4 variables:
## $ counts: int 21 194 259 393 443 117 1123 71 29 5 ...
## $ vote : Factor w/ 2 levels "yes","no": 1 2 1 2 1 2 1 2 1 2 ...
## $ class : Factor w/ 5 levels "High","3rd.current",..: 1 1 2 2 3 3 5 5 4 4 ...
## $ region: Factor w/ 10 levels "Seoul","Yuhu",..: 1 1 1 1 1 1 2 2 3 3 ...
- We add color for the vote.
sejong.poll.2$color[sejong.poll.2$vote=="yes"]<-"cyan"
sejong.poll.2$color[sejong.poll.2$vote=="no"]<-"red"
- Check the total vote with xtabs()
options(digits=3)
xtabs(counts~vote, data=sejong.poll.2)
## vote
## yes no
## 98657 74149
prop.table(xtabs(counts~vote, data=sejong.poll.2))
## vote
## yes no
## 0.571 0.429
- We can check the color. Coordinates of text() are found by locator(2). Try!
pie(xtabs(counts~vote, data=sejong.poll.2), col=sejong.poll.2$color)
title(main="Overall Yes or No")
text(x=0, y=c(0.4,-0.4), labels=c("98657", "74149"))
xtabs(counts~vote+class, data=sejong.poll.2)
## class
## vote High 3rd.current 3rd.former Chief Commons
## yes 21 259 443 192 97742
## no 200 393 117 135 73304
- We need to analyse Commons separately.
sejong.poll.2$class.2<-ifelse(sejong.poll.2$class=="Commons", "Commons", "Bureaus")
str(sejong.poll.2)
## 'data.frame': 44 obs. of 6 variables:
## $ counts : int 21 194 259 393 443 117 1123 71 29 5 ...
## $ vote : Factor w/ 2 levels "yes","no": 1 2 1 2 1 2 1 2 1 2 ...
## $ class : Factor w/ 5 levels "High","3rd.current",..: 1 1 2 2 3 3 5 5 4 4 ...
## $ region : Factor w/ 10 levels "Seoul","Yuhu",..: 1 1 1 1 1 1 2 2 3 3 ...
## $ color : chr "cyan" "red" "cyan" "red" ...
## $ class.2: chr "Bureaus" "Bureaus" "Bureaus" "Bureaus" ...
- Compare the votes by class.2, (Bureaucrats vs Commons)
xtabs(counts~vote+class.2, data=sejong.poll.2)
## class.2
## vote Bureaus Commons
## yes 915 97742
## no 845 73304
- Add subtotals to the margins,
addmargins(xtabs(counts~vote+class.2, data=sejong.poll.2))
## class.2
## vote Bureaus Commons Sum
## yes 915 97742 98657
## no 845 73304 74149
## Sum 1760 171046 172806
- Compute the marginal proportions. Note the use of digits=3.
prop.table(xtabs(counts~vote+class.2, data=sejong.poll.2), margin=2)
## class.2
## vote Bureaus Commons
## yes 0.520 0.571
## no 0.480 0.429
- Pie charts for Bureacrats by vote and Commons by vote.
attach(sejong.poll.2)
par(mfrow=c(1,2))
pie(xtabs(counts~vote+class.2, data=sejong.poll.2[class.2=="Bureaus",], drop=T), labels=c("yes", "no"), col=color)
title(main="Bureacrats by vote")
text(x=0, y=c(0.4,-0.4), labels=c("915", "845"))
pie(xtabs(counts~vote+class.2, data=sejong.poll.2[class.2=="Commons",], drop=T), labels=c("yes", "no"), col=color)
title(main="Commons by vote")
text(x=0, y=c(0.4,-0.4), labels=c("97742", "73304"))
par(mfrow=c(1,1))
- Count the vote by region class.2 wise.
xtabs(counts~vote+region, data=sejong.poll.2[class.2=="Bureaus",], drop=T)
## region
## vote Seoul Gyunggi Pyungan Hwanghae Chungcheong Kangwon Hamgil Gyungsang
## yes 723 29 6 17 35 5 3 55
## no 704 5 36 17 28 10 15 16
## region
## vote Jeolla
## yes 42
## no 14
xtabs(counts~vote+region, data=sejong.poll.2[class.2=="Commons",], drop=T)
## region
## vote Yuhu Gyunggi Pyungan Hwanghae Chungcheong Kangwon Hamgil Gyungsang
## yes 1123 17076 1326 4454 6982 939 75 36262
## no 71 236 28474 15601 14013 6888 7387 377
## region
## vote Jeolla
## yes 29505
## no 257
- Seoul has three times more Bureaucrats than other regions, so analyse further.
xtabs(counts~vote+class, data=sejong.poll.2[region=="Seoul",], drop=T)
## class
## vote High 3rd.current 3rd.former
## yes 21 259 443
## no 194 393 117
- Draw barplot for the vote by class in Seoul. Text positions were obtained by locator().
barplot(xtabs(counts~vote+class, data=sejong.poll.2[region=="Seoul",], drop=T), col=color)
title(main="Seoul by vote")
text(x=c(0.7, 1.9, 1.9, 3.1, 3.1), y=c(120, 450, 135, 500, 220), labels=c("194","393", "259", "117", "443"))
legend("topleft", inset=0.05, fill=c("cyan", "red"), legend=c("yes", "no"))
mosaicplot(xtabs(counts~class+vote, data=sejong.poll.2[region=="Seoul",], drop=T), col=color, main="Seoul by vote")
- Draw barplot() for the Bureaus by region.
xtabs(counts~vote+region, data=sejong.poll.2[class.2=="Bureaus" & !region=="Seoul",], drop=T)
## region
## vote Gyunggi Pyungan Hwanghae Chungcheong Kangwon Hamgil Gyungsang Jeolla
## yes 29 6 17 35 5 3 55 42
## no 5 36 17 28 10 15 16 14
barplot(xtabs(counts~vote+region, data=sejong.poll.2[class.2=="Bureaus" & !region=="Seoul",], drop=T), col=color)
title(main="Bureacrats' vote by region other than Seoul")
legend("topleft", inset=0.05, fill=c("cyan", "red"), legend=c("yes", "no"))
mosaicplot(xtabs(counts~region+vote, data=sejong.poll.2[class.2=="Bureaus" & !region=="Seoul",], drop=T), col=color, main="")
title(main="Bureacrats' vote by region other than Seoul")
- Draw barplot() for the Commons by region.
xtabs(counts~vote+region, data=sejong.poll.2[class.2=="Commons",], drop=T)
## region
## vote Yuhu Gyunggi Pyungan Hwanghae Chungcheong Kangwon Hamgil Gyungsang
## yes 1123 17076 1326 4454 6982 939 75 36262
## no 71 236 28474 15601 14013 6888 7387 377
## region
## vote Jeolla
## yes 29505
## no 257
barplot(xtabs(counts~vote+region, data=sejong.poll.2[class.2=="Commons",], drop=T), col=color)
title(main="Commons' vote by region")
legend("topleft", inset=0.05, fill=c("cyan", "red"), legend=c("yes", "no"))
- Draw by mosaicplot() in base graphics.
mosaicplot(xtabs(counts~region+vote, data=sejong.poll.2[class.2=="Commons",], drop=T), col=color, main="Commons' votes by region")
xtabs(counts~vote+class, data=sejong.poll.2[region=="Chungcheong",], drop=T)
## class
## vote High Chief Commons
## yes 0 35 6982
## no 2 26 14013
prop.table(xtabs(counts~vote+class, data=sejong.poll.2[region=="Chungcheong",], drop=T), margin=2)
## class
## vote High Chief Commons
## yes 0.000 0.574 0.333
## no 1.000 0.426 0.667
barplot(prop.table(xtabs(counts~vote+class, data=sejong.poll.2[region=="Chungcheong",], drop=T), margin=2), col=color, ylim=c(0, 1.5), axes=F)
axis(side=2, at=c(0, 0.5, 1.0), labels=c("0", "50%", "100%"))
title(main="Chungcheong's vote proportion by class")
legend("topleft", inset=0.05, fill=c("cyan", "red"), legend=c("yes", "no"))
text(x=c(0.7, 1.9, 1.9, 3.1, 3.1), y=c(0.5, 0.3, 0.8, 0.15, 0.65), labels=c(2, 35, 26, 6982, 14013))
- With mosaicplot, it’s hard to compare.
mosaicplot(xtabs(counts~class+vote, data=sejong.poll.2[region=="Chungcheong",], drop=T), col=color, main="")
title(main="Chungcheong's vote")
- Save the working directory image, save history and quit.
save.image(file="sejong_poll0328.rda")
savehistory(file="sejong_poll0328.Rhistory")
q("no")