05-Data-Types.Rmd

---
title: "Data Types"
output: html_notebook
---

<!-- This file by Charlotte Wickham is licensed under a Creative Commons Attribution 4.0 International License, adapted from the orignal work at https://github.com/rstudio/master-the-tidyverse by RStudio. -->

```{r setup}
library(tidyverse)
library(lubridate)

# Example of a factor
eyes <- factor(x = c("blue", "green", "green"), 
               levels = c("blue", "brown", "green"))

# An example for times/dates
library(fivethirtyeight)
births <- US_births_1994_2003 %>%
  select(date, births)
```

## Warm-up / Review

Using the data `gss_cat`, find the average hours of tv watched (`tvhours`) for each category of marital status (`marital`).

```{r}
gss_cat
```

## Your Turn 1

What kind of object is the `marital` variable?  

```{r}
gss_cat
```

Brainstorm with your neighbor, all the things you know about that kind of object.

# Factors

## Your Turn 2

Fix your summary of average hours of tv watched (`tvhours`) by marital status (`marital`), to drop missing values in `tvhours`, then create a plot to examine the results.

```{r}
gss_cat %>%
  group_by(marital) %>% 
  summarise(avg_tvhours = mean(tvhours)) 
```

## Your Turn 3

Fill in the blanks (`   `) to explore the average hours of tv watched by religion.

```{r, error = TRUE}
gss_cat %>%
  drop_na(    ) %>%
  group_by(    ) %>%
  summarise(    ) %>%
  ggplot() +
    geom_point(mapping = aes(x =     , y =    ))
```

## Quiz 

Why is this plot not very useful?

```{r}
gss_cat %>%
  drop_na(tvhours) %>%
  group_by(denom) %>%
  summarise(avg_tvhours = mean(tvhours)) %>%
  ggplot() +
    geom_point(mapping = aes(x = avg_tvhours,
      y = fct_reorder(denom, avg_tvhours)))
```

## Your Turn 4

Edit the code to also relabel some other Baptist denominations:

* "Baptist-dk which" -> "Baptist - Don't Know"    
* "Other baptists" -> "Baptist = Other"

```{r}
gss_cat %>%
  mutate(denom = fct_recode(denom,
    "Baptist - Southern" = "Southern baptist")
  ) %>%
  pull(denom) %>%
  levels()
```

## Your Turn 5

What does the function `detect_denom()` do?

```{r}
detect_denom <- function(x){
  case_when(
    str_detect(x, "[Bb]ap") ~ "Baptist", 
    str_detect(x, "[Pp]res") ~ "Presbyterian",
    str_detect(x, "[Ll]uth") ~ "Lutheran",
    str_detect(x, "[Mm]eth") ~ "Methodist",
    TRUE ~ x
  )
}

gss_cat %>% pull(denom) %>% levels() %>% detect_denom()
```

# Strings

With your neighbor, predict what these might return:

```{r}
strings <- c("Apple", "Pineapple", "Orange")

str_detect(strings, pattern = "pp")
str_detect(strings, pattern =  "apple")
str_detect(strings, pattern = "[Aa]pple")
```

Then run them!

# Times and Dates

## Your Turn 7

For each of the following formats (of the same date), pick the right `ymd()` function to parse them:

```{r}
"2018 Feb 01"
"2-1-18"
"01/02/2018"
```

## Your Turn 8

Fill in the blanks to:

* Extract the month from date. 
* Extract the year from date.
* Find the total births for each year/month.
* Plot the results as a line chart.

```{r, error = TRUE}
births %>%
  mutate(year = ___,
    month = ___) %>%
  group_by(___, ___) %>%
  summarise(total_births = ___) %>%
  ggplot() + 
    geom_line(aes(x = month, y = total_births, group = year))
```


# Take Aways

Dplyr gives you three _general_ functions for manipulating data: `mutate()`, `summarise()`, and `group_by()`. Augment these with functions from the packages below, which focus on specific types of data.

Package   | Data Type
--------- | --------
forcats   | factors
stringr   | strings
hms       | times
lubridate | dates and times