week_7.Rmd

---
title: "forcats Tutorial"
author: "Heili Lowman"
date: "9/22/2020"
output: html_document
---

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

## R Markdown "forcats" tutorial

The following document will walk through some functions within the `forcats` package that comes as part of the tidyverse.

### Packages

```{r, message = FALSE}
# Load necessary packages.
library(tidyverse)
```

### Data

```{r, message = FALSE}
# Load necessary data.
#View(trees)

trees_df <- trees # Save data to environment.
```

### Tidy

```{r, message = FALSE}

# Add a column of size classes.
trees_df <- trees_df %>%
  mutate(SizeClass = case_when( # adding in a new column
    Height < 65 ~ "XS",
    Height >= 65 & Height < 70 ~ "S",
    Height >= 70 & Height < 75 ~ "M",
    Height >= 75 & Height < 80 ~ "L",
    Height >= 80 ~ "XL"
  ))

#str(trees_df) # examines structure.

# factor()
trees_df <- trees_df %>%
  mutate(SizeClassF = factor(SizeClass)) # make new factor column
  
#str(trees_df)  

trees_df$SizeClassF <- factor(trees_df$SizeClass) # achieves the same result as the pipe above.

# levels()
levels(trees_df$SizeClassF)

# fct_relevel()
fct_relevel(trees_df$SizeClassF, c ("XS", "S", "M", "L", "XL")) # relevels our categories manually.

trees_df <- trees_df %>%
  mutate(SizeClassF = factor(SizeClass, levels = c("XS", "S", "M", "L", "XL"))) # achieves the same result as using fct_relevel.

#str(trees_df)

```

### Exploration

```{r, message = FALSE}

# fct_count()
count <- fct_count(trees_df$SizeClassF) # Counts number of observations in each level.

# fct_unique()
unique <- fct_unique(trees_df$SizeClassF) # Counts number of unique levels in a dataset.
#View(unique)

# Create a base boxplot of the data.

boxplot1 <- trees_df %>% # Uses original dataset.
  ggplot(aes(x = SizeClassF, y = Height, color = SizeClassF)) + # Sets the axes.
  geom_boxplot() + # Creates a boxplot geometry.
  labs(title = "Base Plot",
    x = "Size Class",
    y = "Height (ft)",
    color = "Size Class",
    caption = "Data Source: Ryan et al., 1976") + # Adds informative labels.
  theme_classic() + # Removes grey background.
  theme(legend.position = "none") # Removes legend.

boxplot1

# fct_infreq()
# Sorts factors in descending order of frequency.

boxplot2 <- trees_df %>%
  mutate(SizeClassF = fct_infreq(SizeClassF)) %>%
  ggplot(aes(x = SizeClassF, y = Height, color = SizeClassF)) +
  geom_boxplot() +
  labs(title = "Frequency Plot",
    x = "Size Class",
    y = "Height (ft)",
    color = "Size Class",
    caption = "Data Source: Ryan et al., 1976") +
  theme_classic() +
  theme(legend.position = "none") # Removes legend.

boxplot2

# fct_reorder()
# Reorders factor column by another column.
# ideally, there should be a 1:1 match between factors and other columns observations.
# fct_reorder2() - allows you to reorder groups based on how their variables plot on a graph (in the x and y directions)

boxplot3 <- trees_df %>%
  mutate(SizeClassF = fct_reorder(SizeClassF, Volume)) %>%
  ggplot(aes(x = SizeClassF, y = Height, color = SizeClassF)) +
  geom_boxplot() +
  labs(title = "Reordered by Volume Plot",
    x = "Size Class",
    y = "Height (ft)",
    color = "Size Class",
    caption = "Data Source: Ryan et al., 1976") +
  theme_classic() +
  theme(legend.position = "none") # Removes legend.

boxplot3

# fct_recode()
# Manually reassign names of categories/factor levels.

boxplot4 <- trees_df %>%
  mutate(SizeClassF = fct_relevel(SizeClassF, c ("XS", "S", "M", "L", "XL"))) %>% # ensures ordering by size class
  mutate(SizeClassF2 = fct_recode(SizeClassF, "X Small" = "XS", "Small" = "S", "Medium" = "M", "Large" = "L", "X Large" = "XL")) %>% # creates a new column with full names, still ordered
  ggplot(aes(x = SizeClassF2, y = Height, color = SizeClassF2)) +
  geom_boxplot() +
  labs(title = "Renamed Plot",
    x = "Size Class",
    y = "Height (ft)",
    color = "Size Class",
    caption = "Data Source: Ryan et al., 1976") +
  theme_classic() +
  theme(legend.position = "none") # Removes legend.

boxplot4

#fct_other()
# Select certain categories, and deem the rest as "other".

boxplot5 <- trees_df %>%
  mutate(SizeClassF3 = fct_other(SizeClassF, keep = c ("S", "M", "L"))) %>% # can also use drop, in this case drop = c("XS", "XL") to achieve an identical result.
  ggplot(aes(x = SizeClassF3, y = Height, color = SizeClassF3)) +
  geom_boxplot() +
  labs(title = "Other Category Plot",
    x = "Size Class",
    y = "Height (ft)",
    color = "Size Class",
    caption = "Data Source: Ryan et al., 1976") +
  theme_classic() +
  theme(legend.position = "none") # Removes legend.

boxplot5

# fct_expand()
# Add a new factor level/category.

trees_expand <- trees_df %>%
  mutate(SizeClassF4 = fct_expand(SizeClassF, "XXL"))

levels(trees_expand$SizeClassF4) # checks levels

# fct_drop()
# Removes a factor/category.
# Do not need to specify factor, because R will recognize those that are unused and drop them.

trees_expand <- trees_expand %>%
  mutate(SizeClassF5 = fct_drop(SizeClassF4))

levels(trees_expand$SizeClassF5)

```

End of RMarkdown document.