05-highlighting-subsets.Rmd

---
title: "05-highlighting with subsets"
output: html_notebook
---

If you've skipped ahead (and not run Part 1), run this code chunk to load the relevant data and plot:

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(here)
source(here("R/init.R"))
```

## Things that will make this section easier to understand

Understanding `dplyr` and how `mutate()` and `filter()` work will help you out a lot in these slides.

If you want to learn the basics of `dplyr`, please check out R-Bootcamp Chapter 3: https://r-bootcamp.netlify.com/chapter3

## Optional: Making a Dummy Variable to color specific lines in our dataset

We can manually color a single group in our data through a two step process. For example, if we want to highlight "Breaking Bad" but not the other shows we can:

1. make a "dummy" variable called `breaking_bad` by recoding the `genre` variable to have `Y` when this variable has "Drama" in it, and `N` for the other categories.

2. manually color the lines using `scale_color_manual()` by specifying a `values()` argument.

In order to make our new variable, we will use a command called `mutate()` to search for "Breaking Bad" in the `title` variable.

### Using a dummy variable in our dataset (highlight a single show)

What if we wanted to highlight a single tv show in the data, such as `Breaking Bad`?

We can create a dummy variable in our dataset, much like `status`. In this one, we want to create a variable called `breaking_bad` where the value is `Y` (yes) where the `title` is `Breaking Bad` and `N` (no) otherwise: 

We can add a new variable using what is called a `mutate()` function. It lets us calculate a new variable, based on the other values of the variable. If you've used Excel's `IF()` function, it's very similar in terms of the syntax. We first provide a condition:

```{r}
tv_shows_breaking <- tv_shows %>%
  mutate(breaking_bad = 
           case_when( 
             title == "Breaking Bad" ~ "Y",
             TRUE ~ "N"
  ))  
```



## Highlighting multiple lines using `case_when`:

```
mutate(
   #check whether a cell has the word "Comedy" in it
      #If it does, output a "Y", 
    comedy = case_when(
        str_detect(genres, "Comedy") ~ "Y", 
      #otherwise, output a "N"
                  TRUE = "N)
      }
```

The `ifelse()` function can be read:

> If the value for `genres` contains "Comedy" THEN recode that value as "Y",
> If it doesn't contain "Comedy", THEN recode that value as "N"

Run the code below and explore the table a little bit to confirm that "Comedy" shows are coded as `Y` in our new variable, `comedy`:

```{r}
tv_shows_new <- tv_shows %>% 
  mutate(
   #check whether a cell has the word "Comedy" in it
      #If it does, output a "Y", 
     comedy = case_when(
        str_detect(genres, "Comedy") ~ "Y", 
      #otherwise, output a "N"
                  TRUE ~ "N"
      )
     )
      

tv_shows_new
```

Now we can use `scale_color_manual()` to remap the `TRUE` and `FALSE` values.

```{r}
    ggplot(tv_shows_new) + 
    aes(x = seasonNumber, y= av_rating, group=title, 
      color = comedy) + 
    geom_line() + 
  scale_color_manual(values=c("Y"="blue", 
                             "N" = "grey")) +
    scale_x_continuous(breaks=c(1:10))
```


## Optional: Highlighting a subset of the data

This is an optional section, but it shows you a specific trick in highlighting your data.

What if we wanted to highlight a specific time period in our dataset? We can actually subset the data, and plot over the original data with a different color to highlight it.

We want to specify this criteria in our `filter()` statement:

```{r}
tv_subset <- tv_shows %>%
  filter(seasonNumber > 2) %>%
  filter(seasonNumber < 6)
```

Now we can use this dataset to highlight this portion of the data:

```{r}

ggplot(tv_shows) + 
    aes(x = seasonNumber, y= av_rating, group=title) + 
  #color all the data - notice we pass a color argument directly in
    geom_line(color = "grey") +
  #color the subset - notice that pass a color argument in directly
  #also notice that we specify `data` to be our `tv_subset`
  geom_line(data = tv_subset, color = "blue") +
  geom_vline(xintercept = 3, lty=3) +
  geom_vline(xintercept = 5, lty=3) +
  annotate("text", x=4, y=7, label="This is a critical\ntime period") +
  labs(title = "Seasons 3 to 5 are Critical") + 
  theme_minimal() + 
  scale_x_continuous(breaks=c(1:10))
```

Now that you know how to use subsets in plotting, you can use this to highlight a group of points as well in a scatterplot as well. Here's an example:

```{r}
high_share_rated <- tv_shows %>% filter(share > 9 & av_rating > 8.5)

ggplot(tv_shows) +
  aes(x = av_rating, y=share) +
  geom_point(color = "grey") +
  geom_point(data = high_share_rated, color = "blue")
```

Here's an example with annotations.

```{r}
high_share_rated <- tv_shows %>% filter(share > 9 & av_rating > 8.5)

ggplot(tv_shows) +
  aes(x = av_rating, y=share) +
  geom_point(color = "grey") +
  geom_point(data = high_share_rated, color = "blue") +
  labs(title = stringr::str_wrap("Only 'Breaking Bad' was high in both ratings and share")) +
  geom_hline(yintercept = 9, lty=3) +
  geom_vline(xintercept = 8.5, lty = 3) +
  #annotate("text", x=9.1, y=15, label="Breaking\nBad") +
  annotate("text", x=9.1, y=18, label= "High Share\nHigh Rating") +
  annotate("text", x=6.5, y=5, label = "Low Share\nLow Rating") +
  theme_minimal()

```