Rachel learns R

library(ggplot2)
Task13 <- read.csv("~/Dropbox/UNDP_HDRO - Internship//Task13 - HALE/HALE.csv", header=T)
Task13$EYS.2013       <- as.numeric(as.character(Task13$EYS.2013))

## Warning: NAs introduced by coercion

Task13$EYS.2014       <- as.numeric(as.character(Task13$EYS.2014))

## Warning: NAs introduced by coercion

Task13$HALE           <- as.numeric(as.character(Task13$HALE))
Task13$HDI.2013       <- as.numeric(as.character(Task13$HDI.2013))

## Warning: NAs introduced by coercion

Task13$HDI.2014       <- as.numeric(as.character(Task13$HDI.2014))

## Warning: NAs introduced by coercion

Task13$Pop.2014       <- as.numeric(as.character(Task13$Pop.2014))
Task13$Pop.2013       <- as.numeric(as.character(Task13$Pop.2013))
Task13$Pop.5.24.2014  <- as.numeric(as.character(Task13$Pop.5.24.2014))
Task13$Pop.5.24.2013  <- as.numeric(as.character(Task13$Pop.5.24.2013))

## Warning: NAs introduced by coercion

Task13$Pop.adult.2014 <- as.numeric(as.character(Task13$Pop.adult.2014))
Task13$Pop.adult.2013 <- as.numeric(as.character(Task13$Pop.adult.2013))
Task13$GNI.2013[Task13$GNI.2013 > 75000] <- 75000
Task13$GNI.2014[Task13$GNI.2014 > 75000] <- 75000
Task13$EYS.2013[Task13$EYS.2013 > 18]    <- 18
Task13$EYS.2014[Task13$EYS.2014 > 18]    <- 18

Show scatterplot of Life Expectancy vs. Health Adjusted Life Expectancy

Reg.col.list <- c("#1b9e77","#d95f02","#7570b3", "#e7298a", "#66a61e", "#e6ab02", "#a6761d", "#666666")
palette(Reg.col.list)
plot(Task13$HALE~Task13$LE.2013,
     main="Health Quartiles in 2013",
     xlab="Life Expectancy 2013",
     ylab="Health-Adjusted Life Expectancy 2013",
     font.lab=2,
     col=Task13$Region,
     #xlim=c(45,85),
     #ylim=c(45,80),
     pch=16,
     cex=.5)

legend(70, 48.5, legend = levels(HALE.diff$Region), col = 1:8, cex = .65, pch = 16)

## Error in levels(HALE.diff$Region): object 'HALE.diff' not found

Histogram of countries with large differences between LE and HALE

ggplot(aes(Task13$LE.2013 - Task13$HALE), data=Task13) +
  ggtitle("Life Expectancy (LE) vs. Health-Adjusted Life Expectancy (HALE):\n2013 Difference") +
  xlab ("Difference: 2013 LE - 2013 HALE") +
  ylab ("Frequency of Difference") +
  theme(plot.title=element_text(face="bold", size=12), axis.ticks=element_blank())+
  geom_bar(aes(fill=Region, labels=TRUE),col="black", binwidth=1) +
  scale_fill_manual(values = c("#1b9e77","#d95f02","#7570b3", "#e7298a","#66a61e", "#e6ab02", "#a6761d", "#666666"))+
  stat_bin(aes(label=..count.., fill=Region, y=..count..), vjust=1.5, geom="text", binwidth=1, colour="black",size=3.5,drop=TRUE) +
  geom_text(stat="bin", binwidth = 1, aes(label=..count..), size=3.75, vjust=-0.25, fontface="bold", drop=TRUE) +
  scale_y_continuous(breaks=seq(0,35,1)) +
  scale_x_continuous(breaks=seq(0,15,1)) +
  annotate("text", label="Angola", cex=3.5, x=-2, y=1.5, fontface="bold") +
  annotate("text", label="Cote\nd'Ivoire", cex=3.5, x=-2, y=.5, fontface="bold",lineheight=.6) +
  annotate("text", label="Sierra\nLeone", cex=3.5, x=.5, y=-.5, fontface="bold",lineheight=.6) +
  annotate("text", label="Turkmenistan", cex=3.5, x=1.75, y=3.5, fontface="bold") +
  annotate("text", label="Malawi", cex=3.5, x=13.5, y=24.5, fontface="bold") +
  annotate("text", label="Namibia", cex=3.5, x=13.6, y=23.75, fontface="bold") +
  annotate("text", label="Bangladesh", cex=3.5, x=14.25, y=22.5, fontface="bold") +
  annotate("text", label="Afghanistan", cex=3.5, x=14.25, y=21.75, fontface="bold") +
  annotate("text", label="Dominica", cex=3, x=12.75, y=-.5, fontface="bold") +
  annotate("text", label="Gabon", cex=3.5, x=14.25, y=10.5, fontface="bold") +
  annotate("text", label="Switzerland", cex=3.5, x=14.75, y=9.5, fontface="bold") +
  annotate("text", label="Australia", cex=3.5, x=14.5, y=8.75, fontface="bold") +
  annotate("text", label="Italy", cex=3.5, x=14, y=8, fontface="bold") +
  annotate("text", label="New Zealand", cex=3.5, x=15, y=7.25, fontface="bold") +
  annotate("text", label="Sweden", cex=3.5, x=14.5, y=6.5, fontface="bold") +
  annotate("text", label="Spain", cex=3.5, x=14.25, y=5.75, fontface="bold") +
  annotate("text", label="Netherlands", cex=3.5, x=15, y=5, fontface="bold") +
  annotate("text", label="Luxembourg", cex=3.5, x=15, y=4.25, fontface="bold") +
  annotate("text", label="Finland", cex=3.5, x=14.5, y=3.5, fontface="bold") +
  annotate("text", label="Vanuatu", cex=3.5, x=15.5, y=.5, fontface="bold") +
  annotate("text", label="Macedonia, Romania, Pakistan, ", cex=3.5, x=2, y=32, fontface="bold") +
  annotate("text", label="Botswana, Congo, Rwanda, Belize, ", cex=3.5, x=2, y=31.5, fontface="bold")+
  annotate("text", label="Syria, Malaysia, Bhutan, Palestine, ", cex=3.5, x=2, y=31, fontface="bold")+
  annotate("text", label="Venezuela, Paraguay, Cabo Verde, ", cex=3.5, x=2.5, y=30.5, fontface="bold")+
  annotate("text", label="Bahrain, Colombia, Algeria, ", cex=3.5, x=2, y=30, fontface="bold")+
  annotate("text", label="Dominican Republic, Papua New Guinea, ", cex=3.5, x=2.5, y=29.5, fontface="bold")+
  annotate("text", label="Bosnia & Herzegovina, Qatar,", cex=3.5, x=2, y=29, fontface="bold")+
  annotate("text", label="Brunei Darussalam, Guinea-Bissau, Hungary,", cex=3.5, x=3, y=28.5, fontface="bold")+
  annotate("text", label="Senegal, Micronesia, Kazakhstan, Mauritius,", cex=3.5, x=3, y=28, fontface="bold")+
  annotate("text", label="Belarus, Samoa, Bulgaria, Croatia", cex=3.5, x=3, y=27.5, fontface="bold")

## ymax not defined: adjusting position using y instead

Show Country Quartiles of 2013 LE and HALE 20

#SCATTERPLOT LE 2013 vs. HALE 2013 CORRELATIONS, color by region with ISO3 labels
palette(Reg.col.list)
plot(Task13$HALE~Task13$LE.2013,
     main="Health Quartiles in 2013",
     xlab="Life Expectancy 2013",
     ylab="Health-Adjusted Life Expectancy 2013",
     font.lab=2,
     col=Task13$Region,
     #xlim=c(45,85),
     #ylim=c(45,80),
     pch=16,
     cex=.5)
legend(70, 48.5, legend = levels(HALE.diff$Region), col = 1:8, cex = .65, pch = 16)

## Error in levels(HALE.diff$Region): object 'HALE.diff' not found

#Create quartile for LOW HEALTH
segments(49.00, 42.10, 49.00, 56.88, col="#525252") #left
segments(49.00, 42.10, 64.88, 42.10, col="#525252") #bottom
segments(64.88, 42.10, 64.88, 56.88, col="#525252") #right
segments(49.00, 56.88, 64.88, 56.88, col="#525252") #top

#Create quartile for MED HDI
segments(64.88, 56.88, 64.88, 64.05, col="#525252") #left
segments(64.88, 56.88, 72.95, 56.88, col="#525252") #bottom
segments(72.95, 56.88, 72.95, 64.05,col="#525252") #right
segments(64.88, 64.05, 72.95, 64.05,col="#525252") #top

#Create quartile for HIGH HDI
segments(72.95, 64.05, 72.95, 67.38, col="#525252") #left
segments(72.95, 64.05, 76.60, 64.05 , col="#525252") #bottom
segments(76.60, 64.05, 76.60, 67.38, col="#525252") #right
segments(72.95, 67.38, 76.60, 67.38, col="#525252") #top

#Create quartile for VERY HIGH HDI
segments(76.60, 67.38, 76.60, 73.38, col="#525252") #left
segments(76.60, 67.38, 83.80, 67.38 , col="#525252") #bottom  
segments(83.80, 67.38, 83.80, 73.38, col="#525252") #right  
segments(76.60, 73.38, 83.80, 73.38, col="#525252") #top  

text(x = 56, y = 60, cex = .7, labels = "Low LE\n Medium HALE")
text(x = 66.5, y = 65.4, cex = .7, labels = "Medium LE\n High HALE")
text(x = 73.5, y = 72, cex = .7, labels = "High LE\n Very High HALE")
text(x = 68, y = 52, cex = .7, labels = "Medium LE\n Low HALE")
text(x = 76, y = 60, cex = .7, labels = "High LE\n Medium HALE")
text(x = 82, y = 65, cex = .65, labels = "Very High LE\n High HALE")

ggplot(aes(Task13$LE.2013 - Task13$HALE), data=Task13) +
  ggtitle("Life Expectancy (LE) vs. Health-Adjusted Life Expectancy (HALE):\n2013 Difference") +
  xlab ("Difference: 2013 LE - 2013 HALE") +
  ylab ("Frequency of Difference") +
  theme(plot.title=element_text(face="bold", size=12), axis.ticks=element_blank())+
  geom_bar(aes(fill=Region, labels=TRUE),col="black", binwidth=1) +
  scale_fill_manual(values = c("#1b9e77","#d95f02","#7570b3", "#e7298a","#66a61e", "#e6ab02", "#a6761d", "#666666"))+
  stat_bin(aes(label=..count.., fill=Region, y=..count..), vjust=1.5, geom="text", binwidth=1, colour="black",size=3.5,drop=TRUE) +
  geom_text(stat="bin", binwidth = 1, aes(label=..count..), size=3.75, vjust=-0.25, fontface="bold", drop=TRUE) +
  scale_y_continuous(breaks=seq(0,35,1)) +
  scale_x_continuous(breaks=seq(0,15,1)) +
  annotate("text", label="Angola", cex=3.5, x=-2, y=1.5, fontface="bold") +
  annotate("text", label="Cote\nd'Ivoire", cex=3.5, x=-2, y=.5, fontface="bold",lineheight=.6) +
  annotate("text", label="Sierra\nLeone", cex=3.5, x=.5, y=-.5, fontface="bold",lineheight=.6) +
  annotate("text", label="Turkmenistan", cex=3.5, x=1.75, y=3.5, fontface="bold") +
  annotate("text", label="Malawi", cex=3.5, x=13.5, y=24.5, fontface="bold") +
  annotate("text", label="Namibia", cex=3.5, x=13.6, y=23.75, fontface="bold") +
  annotate("text", label="Bangladesh", cex=3.5, x=14.25, y=22.5, fontface="bold") +
  annotate("text", label="Afghanistan", cex=3.5, x=14.25, y=21.75, fontface="bold") +
  annotate("text", label="Dominica", cex=3, x=12.75, y=-.5, fontface="bold") +
  annotate("text", label="Gabon", cex=3.5, x=14.25, y=10.5, fontface="bold") +
  annotate("text", label="Switzerland", cex=3.5, x=14.75, y=9.5, fontface="bold") +
  annotate("text", label="Australia", cex=3.5, x=14.5, y=8.75, fontface="bold") +
  annotate("text", label="Italy", cex=3.5, x=14, y=8, fontface="bold") +
  annotate("text", label="New Zealand", cex=3.5, x=15, y=7.25, fontface="bold") +
  annotate("text", label="Sweden", cex=3.5, x=14.5, y=6.5, fontface="bold") +
  annotate("text", label="Spain", cex=3.5, x=14.25, y=5.75, fontface="bold") +
  annotate("text", label="Netherlands", cex=3.5, x=15, y=5, fontface="bold") +
  annotate("text", label="Luxembourg", cex=3.5, x=15, y=4.25, fontface="bold") +
  annotate("text", label="Finland", cex=3.5, x=14.5, y=3.5, fontface="bold") +
  annotate("text", label="Vanuatu", cex=3.5, x=15.5, y=.5, fontface="bold") +
  annotate("text", label="Macedonia, Romania, Pakistan, ", cex=3.5, x=2, y=32, fontface="bold") +
  annotate("text", label="Botswana, Congo, Rwanda, Belize, ", cex=3.5, x=2, y=31.5, fontface="bold")+
  annotate("text", label="Syria, Malaysia, Bhutan, Palestine, ", cex=3.5, x=2, y=31, fontface="bold")+
  annotate("text", label="Venezuela, Paraguay, Cabo Verde, ", cex=3.5, x=2.5, y=30.5, fontface="bold")+
  annotate("text", label="Bahrain, Colombia, Algeria, ", cex=3.5, x=2, y=30, fontface="bold")+
  annotate("text", label="Dominican Republic, Papua New Guinea, ", cex=3.5, x=2.5, y=29.5, fontface="bold")+
  annotate("text", label="Bosnia & Herzegovina, Qatar,", cex=3.5, x=2, y=29, fontface="bold")+
  annotate("text", label="Brunei Darussalam, Guinea-Bissau, Hungary,", cex=3.5, x=3, y=28.5, fontface="bold")+
  annotate("text", label="Senegal, Micronesia, Kazakhstan, Mauritius,", cex=3.5, x=3, y=28, fontface="bold")+
  annotate("text", label="Belarus, Samoa, Bulgaria, Croatia", cex=3.5, x=3, y=27.5, fontface="bold")

## ymax not defined: adjusting position using y instead

Set up my data

Show scatterplot of Life Expectancy vs. Health Adjusted Life Expectancy

Histogram of countries with large differences between LE and HALE

Show Country Quartiles of 2013 LE and HALE 20