Part2-workingWithFactors.Rmd

---
title: "Part 2 - Working with Factors"
author: "Ted Laderas (Solutions)"
date: "5/18/2017"
output: 
  html_document:
    code_download: true
    code_folding: hide
    df_print: paged
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, message = FALSE)
```

In this section, we'll learn some skills about manipulating factor (categorical) data.

We'll do this by making a bar plot and a box plot and progressively making it more complex.

## Reviewing Factors 

Factors are how R represents categorical data.

There are two kinds of factors: 

+ `factor` - used for *nominal* data ("Ducks","Cats","Dogs")
+ `ordered`- used for *ordinal* data ("10-30","31-40","41-60")

We'll manipulate our barplot and add more information using factors.

Here's the simple dataset we'll use to investigate how to work with factors in `ggplot`.

For the `factor` and `ordered` variables, find the categories for each using `levels()`.

```{r}
library(tidyverse)
load("data/pets.rda")
pets
```

## A Basic Barplot using `geom_bar()`

The `geom_bar()` default is to count the number of values with each factor level. Note that you don't map to a y-aesthetic here, because the y values are the counts.

Given this dataset, we might want to ask how many pets have the same name. What is the most popular name?

Try mapping another variable to `fill` (try both `weight` and `animal`). Are the results what you expected?

```{r}
ggplot(data=pets, mapping=aes(x=name)) + geom_bar()
```

## Faceting

Say you have another `factor` variable and you want to stratify the plots based on that. 
You can do that by supplying the name of that variable as a facet. Here, we facet our barplot by `shotsCurrent`.

```{r}
ggplot(data=pets, mapping=aes(x=name)) + geom_bar() + 
  ## have to specify facets using notation
  ## try out facets=~ageCategory!
  facet_wrap(facets=~shotsCurrent) + 
  ## we make the x axis x angled for better legibility
  theme(axis.text.x = element_text(angle=45))
```

You might notice that there are blank spots for the categories in each facet. We can restrict the factors in each by using `scale="free_x"` argument in `facet_wrap()`.

How many animals named "Morris" did not receive shots?

What happens when you replace the `scale` argument with "free_y"?

```{r}
ggplot(pets, aes(x=name)) + geom_bar() + 
  facet_wrap(facets=~shotsCurrent, scale="free_x") +
  theme(axis.text.x = element_text(angle=45))
```

## Stacked Bars

Let's see how many of each animal got shots. We can do this by mapping `shotsCurrent` to `fill`.

```{r}
#we map color (the outline of the plot) to black to make it look prettier
ggplot(pets, aes(x=animal, fill=shotsCurrent)) + 
  geom_bar(color="black")
```

## Proportional Barchart

We may only be interested in the relative proportions between the different categories. Visualizing this is useful for various 2 x 2 tests on proportions.

What percent of dogs did not receive shots?

```{r}
ggplot(pets, aes(x=animal,fill=shotsCurrent)) + 
  geom_bar(position = "fill", color = "black")
```

## Dodge those bars!

Instead of stacking, we can also dodge the bars (move the bars so they're beside each other).

```{r}
ggplot(pets, aes(x=animal,fill=shotsCurrent)) + 
  geom_bar(position="dodge", color="black")
```

## Your Task: Bar Charts

Given the `pets` `data.frame`, plot a stacked proportional barchart that shows the age category counts by animal type. Is the proportion of animals receiving shots the same across each age category?

Hints: think about what to map to `x`, and what to map to `fill`.

Intermediate Folks: facet this plot by `shotsCurrent`. 

```{r}
#Space for your answer here.

```

```{r}
#Space for your answer here.

```


## Boxplots

Boxplots allow us to assess distributions of a continuous variable conditioned on categorical variables.

What does this tell us? 

```{r}
ggplot(pets, aes(x=shotsCurrent, y=weight)) + geom_boxplot()
```

## Violin Plots

Violin plots are another useful way to visualize the data. They provide a more nuanced look at the data. They're a density plot that's mirrored around the y-axis.

```{r}
ggplot(pets, aes(x=ageCategory, y=weight, fill=ageCategory)) + geom_violin()
```

## Your task: How heavy are our pets?

Visualize `weight` by `animal` type as both a boxplot and a violin plot. What do you conclude? Which kind of animal weighs more on average than the other?

Intermediate challenge: How would we plot both boxplots and a violin plot on the same graph?

```{r}
##Space for your answer here

```

## What you learned in this section

- Visualizing factor data
- Simple, stacked, stacked proportional, and dodged barplots 
- Faceting a graph
- Boxplots and violin plots