cm10.rmd

Class meeting 10, Wednesday Oct 08
========================================================

Reordering factors rationally. Example of the benefit: cause the data and the color guide to be in same order

  * good example: Mohammadreza Bolandnazar: [source](https://gist.github.com/ArephB/6866827#file-stat545a-2013-hw05_bolandnazar-moh-rmd) | [report](http://rpubs.com/aref/stat545a-2013-hw05_bolandnazar-moh)
    - earnings over time by province and gender
    - kudos for reordering the provinces based on earnings in the stripplot
    - go look at Matt Gingerich's work to see how to rotate the x axis tick labels: [source](https://gist.github.com/MattGingerich/6866066#file-stat545a-2013-hw05_gingerich-mat-rmd) | [report](http://rpubs.com/majugi/stat545a-2013-hw05_gingerich-mat) 
  * good example: Rebecca Johnston [source](https://gist.github.com/rebjoh/6870958#file-stat545a-2013-hw05_johnston-reb-rmd) | [report](http://rpubs.com/rljohn/stat545a-2013-hw05_johnston-reb)
    - max # killed barchart by region: rotate? nice reordering
  * example of what we're trying to fix: gao-wen [source](https://gist.github.com/sibyl229/6864753#file-stat545a-2013-hw05_gao-wen-rmd) | [report](http://rpubs.com/less/stat545a-2013-hw05_gao-wen)
    - nice `ggplot2` stripplot of lifeExp ~ year w/ continent-specific loess overlaid; can you get the legend order to better match data order?

```{r include = FALSE}
gdURL <- "http://www.stat.ubc.ca/~jenny/notOcto/STAT545A/examples/gapminder/data/gapminderDataFiveYear.txt"
gDat <- read.delim(file = gdURL)
gDat <- droplevels(subset(gDat, continent != "Oceania"))
library(ggplot2)
```
```{r fig.width = 8, fig.height = 4}
pExp <- ggplot(gDat, aes(x = year, y = lifeExp,
                         color = reorder(continent, -1 * lifeExp)))
pExp + geom_point(alpha=0.3) + geom_smooth(method='loess')
## I obviously need to rename the legend
```

Dean Attali: [source](https://gist.github.com/daattali/6847430#file-stat545a-2013-hw05_attali-dea-rmd) | [report](http://rpubs.com/daattali/stat545a-2013-hw05_attali-dea)

  - terrorist incidents in Israel
  - [How pie charts fail](http://blog.revolutionanalytics.com/2009/08/how-pie-charts-fail.html) from the Revolution Analytics blog
  - pie chart partially redeemed by reordering the factor first
  - `scale_fill_brewer()` gives direct access to `RColorBrewer` palettes
  - reading his code is like reading English, it's just a smooth experience
    
Rebecca Johnston [source](https://gist.github.com/rebjoh/6870958#file-stat545a-2013-hw05_johnston-reb-rmd) | [report](http://rpubs.com/rljohn/stat545a-2013-hw05_johnston-reb)

  - Global Terrorism Database; 105K observations ... maybe same source as attali used?
  - ah, yes! how could I not have mentioned `na.rm` yet?
  - talk about empty facets, NAs, blanks, etc.
  - talk about long factor levels

`ggplot2` legends and their (seeming?) inflexibility

  * has anyone experimented with changing location and shape of the legend? i.e. at the top, spread out wide, etc.?
  * Yumian asked good question: howto put the continent color key on a `ggplot2`-generated graphic? see [this](block10_latticeNittyGritty.html#student-q-a) for example of what we mean