08-List-Columns.Rmd

---
title: "List-Columns"
output: html_notebook
---

```{r setup, message=FALSE}
library(tidyverse)
library(broom)
library(gapminder)
library(stringr)

knitr::opts_chunk$set(warning = FALSE, message = FALSE)
```

## Your Turn 1

How has life expectancy changed since 1952?

Using `gapminder`, make a line plot of **lifeExp** vs. **year** grouped by **country**. Set alpha to 0.2, to see the results better.

```{r}

```

## Your Turn 2

1. Group `gapminder` by `country` and `continent` then fit a model and collect the residuals for *each country.*

2. Plot the residuals vs `year` as a line graph, grouped by `country`, with `alpha = 0.2`.

3. Add the following to your plot to facet by `continent`:

    + facet_wrap(~continent)

```{r warning = FALSE}

```

## Your Turn 3

Complete the code to filter, `residuals`, which is the dataset that you made in Your Turn 2, against `bad_fits`, which is the data set that contains just the countries that have an r-squared < 0.25.

Use the result to plot a line graph of `year` vs. `.resid` colored by `country` for each country that had an r-squared < 0.25.

```{r warning = FALSE}
bad_fits <- gapminder %>% 
  group_by(_______) %>% 
  do(lm(lifeExp ~ year, data = ____) %>%  glance()) %>% 
  filter(___________) 

residuals <- gapminder %>% 
  group_by(_______) %>% 
  do(________________________________________) 

residuals %>% 
  ___________(bad_fits) %>% 
  ggplot() +
    _________________________________________
```

## Your Turn 4

Create your own copy of master and then add one more list column: 

* **output** which contains the output of `augment()` for each model

```{r warning = FALSE}
fit_model <- function(df) lm(lifeExp ~ year, data = df)
get_rsq <- function(mod) glance(mod)$r.squared
_______________________________________________ # write a function that applies augment ot a model

master <- gapminder %>% 
  ___________ %>% # group gapminder
  ___________ %>% # nest gapminder
  mutate(model =  ________________________, # add a model column
         r.squared = _____________________, # add an r.squared column
         output = ________________________) # add an output column
master
```

## Your Turn 5

Use master to recreate our plot of the residuals vs `year` for the six countries with an r squared less than 0.25.

```{r}
  
```

***

# Take Aways

* A two way table is an organizational device that you can manipulate.   
* The table structure will maintain correspondence between the table contents during your manipulations.  
* Data frames can store more than values. They can store list columns, which can contain _any_ type of R object.  
* You can manipulate list columns with `mutate()` and the `map()` family functions.  
* Create list columns with `nest()`  
* Unnest list columns with `unnest()` to create plots  
* Dplyr's `do()` function can be a useful tool when combined with broom functions and `group_by()`