index.Rmd

---
title       : Plotting predictors
subtitle    : 
author      : Jeffrey Leek
job         : Johns Hopkins Bloomberg School of Public Health
logo        : bloomberg_shield.png
framework   : io2012        # {io2012, html5slides, shower, dzslides, ...}
highlighter : highlight.js  # {highlight.js, prettify, highlight}
hitheme     : tomorrow   # 
url:
  lib: ../../librariesNew
  assets: ../../assets
widgets     : [mathjax]            # {mathjax, quiz, bootstrap}
mode        : selfcontained # {standalone, draft}
---


```{r setup, cache = F, echo = F, message = F, warning = F, tidy = F}
# make this an external chunk that can be included in any file
options(width = 100)
opts_chunk$set(message = F, error = F, warning = F, comment = NA, fig.align = 'center', dpi = 100, tidy = F, cache.path = '.cache/', fig.path = 'fig/')

options(xtable.type = 'html')
knit_hooks$set(inline = function(x) {
  if(is.numeric(x)) {
    round(x, getOption('digits'))
  } else {
    paste(as.character(x), collapse = ', ')
  }
})
knit_hooks$set(plot = knitr:::hook_plot_html)
```


## Example: predicting wages

<img class=center src=../../assets/img/08_PredictionAndMachineLearning/wages.jpg height=350>

Image Credit [http://www.cahs-media.org/the-high-cost-of-low-wages](http://www.cahs-media.org/the-high-cost-of-low-wages)

Data from: [ISLR package](http://cran.r-project.org/web/packages/ISLR) from the book: [Introduction to statistical learning](http://www-bcf.usc.edu/~gareth/ISL/)  



---

## Example: Wage data

```{r loadData,cache=TRUE}
library(ISLR); library(ggplot2); library(caret); library(gridExtra);
data(Wage)
summary(Wage)
```



---

## Get training/test sets

```{r trainingTest,dependson="loadData",cache=TRUE}
inTrain <- createDataPartition(y=Wage$wage,
                              p=0.7, list=FALSE)
training <- Wage[inTrain,]
testing <- Wage[-inTrain,]
dim(training); dim(testing)
```


---

## Feature plot (*caret* package)

```{r ,dependson="trainingTest",fig.height=4,fig.width=4}
featurePlot(x=training[,c("age","education","jobclass")],
            y = training$wage,
            plot="pairs")
```


---

## Qplot (*ggplot2* package)


```{r ,dependson="trainingTest",fig.height=4,fig.width=6}
qplot(age,wage,data=training)
```


---

## Qplot with color (*ggplot2* package)


```{r ,dependson="trainingTest",fig.height=4,fig.width=6}
qplot(age,wage,colour=jobclass,data=training)
```


---

## Add regression smoothers (*ggplot2* package)


```{r ,dependson="trainingTest",fig.height=4,fig.width=6}
qq <- qplot(age,wage,colour=education,data=training)
qq +  geom_smooth(method='lm',formula=y~x)
```


---

## cut2, making factors (*Hmisc* package)


```{r cut2,dependson="trainingTest",fig.height=4,fig.width=6,cache=TRUE}
cutWage <- cut2(training$wage,g=3)
table(cutWage)
```

---

## Boxplots with cut2


```{r ,dependson="cut2plot",fig.height=4,fig.width=6,cache=TRUE}
p1 <- qplot(cutWage,age, data=training,fill=cutWage,
      geom=c("boxplot"))
p1
```

---

## Boxplots with points overlayed


```{r ,dependson="cut2plot",fig.height=4,fig.width=9}
p2 <- qplot(cutWage,age, data=training,fill=cutWage,
      geom=c("boxplot","jitter"))
grid.arrange(p1,p2,ncol=2)
```


---

## Tables

```{r ,dependson="cut2",fig.height=4,fig.width=9}
t1 <- table(cutWage,training$jobclass)
t1
prop.table(t1,1)
```


---

## Density plots

```{r ,dependson="trainingTest",fig.height=4,fig.width=6}
qplot(wage,colour=education,data=training,geom="density")
```


---

## Notes and further reading

* Make your plots only in the training set 
  * Don't use the test set for exploration!
* Things you should be looking for
  * Imbalance in outcomes/predictors
  * Outliers 
  * Groups of points not explained by a predictor
  * Skewed variables 
* [ggplot2 tutorial](http://rstudio-pubs-static.s3.amazonaws.com/2176_75884214fc524dc0bc2a140573da38bb.html)
* [caret visualizations](http://caret.r-forge.r-project.org/visualizations.html)