Stat545A Homework #3 Visualize a quantitative variable with lattice

In this homework we will examine the result of the data aggregation task graphically with lattice and ggplot2 . I will use Jenny's solutions as a template for the graphical analysis.

Since next week's homework is to use ggplot2 (and I am more familiar with it now) I will just make both lattice and ggplot2 figures in one go :)

Let's begin:

Basic import and quality control

library(plyr)
library(xtable)
library(lattice)
library(ggplot2)
gDat <- read.delim("gapminderDataFiveYear.txt")
str(gDat)

## 'data.frame':    1704 obs. of  6 variables:
##  $ country  : Factor w/ 142 levels "Afghanistan",..: 1 1 1 1 1 1 1 1 1 1 ...
##  $ year     : int  1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 ...
##  $ pop      : num  8425333 9240934 10267083 11537966 13079460 ...
##  $ continent: Factor w/ 5 levels "Africa","Americas",..: 3 3 3 3 3 3 3 3 3 3 ...
##  $ lifeExp  : num  28.8 30.3 32 34 36.1 ...
##  $ gdpPercap: num  779 821 853 836 740 ...

tail(gDat)

##       country year      pop continent lifeExp gdpPercap
## 1699 Zimbabwe 1982  7636524    Africa   60.36     788.9
## 1700 Zimbabwe 1987  9216418    Africa   62.35     706.2
## 1701 Zimbabwe 1992 10704340    Africa   60.38     693.4
## 1702 Zimbabwe 1997 11404948    Africa   46.81     792.4
## 1703 Zimbabwe 2002 11926563    Africa   39.99     672.0
## 1704 Zimbabwe 2007 12311143    Africa   43.49     469.7

I will use Jenny's htmlPrint table to make the printing process a little easier

htmlPrint <- function(x, ...,
   digits = 0, include.rownames = FALSE) {
   print(xtable(x, digits = digits, ...), type = 'html',
         include.rownames = include.rownames, ...)
   }

Get rid of Oceania

No disrespect to our commonwealth folks down under, but for practical purposes a continent with only two countries is not that interesting. Let's remove it from here on out:

gDat <- droplevels(subset(gDat, continent != "Oceania"))

Easy analysis of GDP “trends”

An easy first plot would be to look at the min and max GDP per capita within the continents over time. Since we are going to graph these data we want a tall format (and need to modify JB's code to include all dates)

foo <- ddply(gDat, .(continent, year), function(x) {
  gdpPercap <- range(x$gdpPercap)
  return(data.frame(gdpPercap, stat = c("min", "max")))
  })
htmlPrint(foo)

Here are the first two interesting plots broken down into continent changes over time.

xyplot(gdpPercap ~ year | continent, data = foo, groups = stat, auto.key = T, type = "o")

ggplot(foo, aes(x = year, y = gdpPercap, color = stat)) + geom_line() +
geom_point(shape = 1) + facet_wrap(~continent)

This probably looks better if we choose a more informative grouping. We should show all of the max on the same scale and then the min on the same scale.

xyplot(gdpPercap ~ year | stat, data = foo, groups = continent, auto.key = T, type = "o")

ggplot(foo, aes(x = year, y = gdpPercap, color = continent)) +
  geom_line() + geom_point(shape = 1) + facet_wrap(~stat)

Two things are annoying here:

• The scales should be different between the two plots (we might need to avoid the faceting and submit to making two different plots)
• The factors are in the wrong order…However, this is a bit of an interesting part of the factor labeling because the ordering is not invariant across min and max (as well as year).

xyplot(gdpPercap ~ year | stat, data = foo, groups = continent, auto.key = T, scales = "free", type = "o")

ggplot(foo, aes(x = year, y = gdpPercap, color = continent)) +
  geom_line() + geom_point(shape = 1) +  facet_grid(stat ~ ., scales = "free")

This solves one problem, but not the other…I don't have a good solution for labeling when the factors are not invariant across groupings.

• What happened to Asia from the 70's to the 80's? It is pretty obvious that the max GDP per capita continued to fall during this period of time.

Absolute and relative world population in each of the continents

The second task I will explore is from Dean's report. He has looked at the absolute and relative world population in each of the continents.

worldRelativePop <- ddply(gDat, .(continent, year),
  function(.data) {
    .data <- as.list(.data)
    .data['continentPop'] <- sum(.data$pop)
    .data['worldPop'] <- sum(subset(gDat, year == .data$year[1])[['pop']])
    .data['percent'] <- round(as.numeric(.data['continentPop'])/as.numeric(.data['worldPop'])*100,2) 
    quickdf(.data[c("continentPop","worldPop",'percent')])
  }
)
htmlPrint(worldRelativePop, digits = c(0,0,0,0,0,2))

First thing I noticed was the use of this cool function quickdf which is a plyr implementation of creating a dataframe quickly. I then noticed that I was not getting the same results as Dean; however, the difference is due to the removal of Oceania! I will process without further worry after this check.

Let us first see how population changes by continent over time:

xyplot(continentPop ~ year, data = worldRelativePop, groups = continent, auto.key = T, type = c("o", "p"))

ggplot(worldRelativePop, aes(year, continentPop, color = continent)) + geom_line() + geom_point()

That was fairly painless, but the legend needs to be fixed. This seems to be more annoying/difficult than originally anticipated since the ordering needs to be done across some function of the chart in the time series as opposed to just one value. Need to ask JB about the proper approach for this functionality….

• More interesting would be to see how the continental population percentages change through time:

xyplot(percent ~ year, data = worldRelativePop, groups = continent, auto.key = T, type = c("o", "p"))

ggplot(worldRelativePop, aes(year, percent, color = continent)) + geom_line() + geom_point()

Clearly Asia is the winner, the more interesting details are in the other continents:

foo <- droplevels(subset(worldRelativePop, continent != "Asia"))
xyplot(percent ~ year, data = foo, groups = continent, auto.key = T, type = c("o", "p"))

ggplot(foo, aes(year, percent, color = continent)) + geom_line() + geom_point()

Now we can clearly see some interesting trends:

• Americas population as a percent of the world's total population has not dramatically changed through time
• The increase in Africa's relative population is in line with Europe's demise in population

Another point of interest would be to see how stationary these percent distributions are (and I want to try a density plot in ggplot2). But, first we will do some stripplots

stripplot(percent ~ reorder(continent, percent), jitter.data = T, worldRelativePop, type = c("p", "a"))

ggplot(worldRelativePop) + geom_point(position = position_jitter(width = 0.15), aes(reorder(continent, percent), percent), shape = 1)

densityplot(~percent | continent, data = worldRelativePop, scales = "free")

ggplot(worldRelativePop, aes(percent)) + stat_density(geom = "path", position = "identity") +
  facet_wrap(~continent, scales = "free")

ggplot does not actually do a good good reproducing these figures from lattice…I think there is likely a way to fine tune the grid over where the kernel estimate is being calculated, although I do not have time to look through too much documentation now.

####

date()

## [1] "Sun Sep 29 23:08:15 2013"

sessionInfo()

## R version 3.0.1 (2013-05-16)
## Platform: x86_64-apple-darwin10.8.0 (64-bit)
## 
## locale:
## [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
## [1] ggplot2_0.9.3.1 lattice_0.20-23 xtable_1.7-1    plyr_1.8       
## [5] knitr_1.4.1    
## 
## loaded via a namespace (and not attached):
##  [1] colorspace_1.2-2   dichromat_2.0-0    digest_0.6.3      
##  [4] evaluate_0.4.7     formatR_0.9        grid_3.0.1        
##  [7] gtable_0.1.2       labeling_0.2       MASS_7.3-26       
## [10] munsell_0.4.2      proto_0.3-10       RColorBrewer_1.0-5
## [13] reshape2_1.2.2     scales_0.2.3       stringr_0.6.2     
## [16] tools_3.0.1