---
title: "Analyzing Hass USA"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Analyzing Hass USA}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

The {avocado} package provides a weekly summary - starting from January 2017 through November 2020 - of Hass Avocado sales. There are three datasets in this package and let's start with the dataset `hass_usa` which focuses on weekly avocado sales in the contiguous US.

Let's start by loading the package - along with {dplyr} (for data wrangling) and {ggplot} (for data visualization) - and exploring it's structure

```{r setup}
library(avocado)
library(dplyr)
library(ggplot2)

data('hass_usa')

dplyr::glimpse(hass_usa)
```


## Exploratory Data Analysis

Let's begin by exploring the following two topics:

- Fluctuation of average selling price
- Non Organic Avocado sales revenue

### Fluctuation of Average Selling Price

```{r fig.height=5, fig.width=8}
hass_usa %>%
  ggplot(aes(x = week_ending)) +
  geom_line(aes(y = avg_price_nonorg, color = 'Non Organic')) +
  geom_line(aes(y = avg_price_org, color = 'Organic')) +
  scale_color_manual(values = c('steelblue','forestgreen')) +
  labs(x = 'Year', y = 'Average Price per Pound in US$', color = 'Type', title = 'Average Selling Price by Week', caption = 'Not adjusted for inflation\nSource: Hass Avocado Board') +
  ylim(min = 0, max = 3.0) +
  theme(plot.background = element_rect(fill = "grey20"),
        plot.title = element_text(color = "#FFFFFF"),
        axis.title = element_text(color = "#FFFFFF"),
        axis.text.x = element_text(color = 'grey50', angle = 45, hjust = 1),
        axis.text.y = element_text(color = 'grey50'),
        plot.caption = element_text(color = 'grey75'),
        panel.background = element_blank(),
        panel.grid.major = element_line(color = "grey50", size = 0.2),
        panel.grid.minor = element_line(color = "grey50", size = 0.2),
        legend.background = element_rect(fill = 'grey20'),
        legend.key = element_rect(fill = 'grey20'),
        legend.title = element_text(color = 'grey75'),
        legend.text = element_text(color = 'grey75'),
        legend.position = c(0.815, 0.85)
        )
```

Unsurprisingly, we can see that the average selling price for organic avocados tends to be higher than the average selling price for non-organic avocados. Note how there seems to be a fairly large spike in selling price in late 2017.

### Non Organic Avocado Sales Revenue

```{r fig.height=5, fig.width=8}
hass_usa %>%
  mutate(
    nonorg_rev = (plu4046+plu4225+plu4770+small_nonorg_bag+large_nonorg_bag+xlarge_nonorg_bag) * avg_price_nonorg
  ) %>%
  ggplot(aes(x = week_ending)) +
  geom_line(aes(y = nonorg_rev/1000000), color = 'steelblue') +
  labs(x = 'Year', y = 'Revenue (Millions US$)', color = 'Type', title = 'Non Organic Avocado Sales Revenue', caption = 'Not adjusted for inflation\nSource: Hass Avocado Board') +
  theme(plot.background = element_rect(fill = "grey20"),
        plot.title = element_text(color = "#FFFFFF"),
        axis.title = element_text(color = "#FFFFFF"),
        axis.text.x = element_text(color = 'grey50'),
        axis.text.y = element_text(color = 'grey50'),
        plot.caption = element_text(color = 'grey75'),
        panel.background = element_blank(),
        panel.grid.major = element_line(color = "grey50", size = 0.2),
        panel.grid.minor = element_line(color = "grey50", size = 0.2),
        legend.background = element_rect(fill = 'grey20'),
        legend.key = element_rect(fill = 'grey20'),
        legend.title = element_text(color = 'grey75'),
        legend.text = element_text(color = 'grey75'),
        # legend.position = c(0.815, 0.2)
        legend.position = 'none'
  )
```

We may have switched gears from looking at average selling price to revenue. However, notice the massive drop in revenue in late 2017 - which seems to correspond to the increase in average selling price for the same time period.

Let's dive further into that 'dip'.

```{r fig.height=5, fig.width=8}
hass_usa %>%
  mutate(
    nonorg_rev = (plu4046+plu4225+plu4770+small_nonorg_bag+large_nonorg_bag+xlarge_nonorg_bag) * avg_price_nonorg
  ) %>%
  select(week_ending, nonorg_rev) %>%
  filter(week_ending > lubridate::ymd('2017-09-10') & week_ending < lubridate::ymd('2017-10-15')) %>%
  ggplot(aes(x = week_ending, y = nonorg_rev)) +
  geom_line(color = 'steelblue') +
  labs(x = 'Year', y = 'Revenue', color = 'Type', title = 'Non Organic Avocado Sales Revenue', caption = 'Not adjusted for inflation\nSource: Hass Avocado Board') +
  theme(plot.background = element_rect(fill = "grey20"),
    plot.title = element_text(color = "#FFFFFF"),
    axis.title = element_text(color = "#FFFFFF"),
    axis.text.x = element_text(color = 'grey50'),
    axis.text.y = element_text(color = 'grey50'),
    plot.caption = element_text(color = 'grey75'),
    panel.background = element_blank(),
    panel.grid.major = element_line(color = "grey50", size = 0.2),
    panel.grid.minor = element_line(color = "grey50", size = 0.2),
    legend.background = element_rect(fill = 'grey20'),
    legend.key = element_rect(fill = 'grey20'),
    legend.title = element_text(color = 'grey75'),
    legend.text = element_text(color = 'grey75'),
    legend.position = 'none'
  )
```

So we can see that the revenue wasn't exactly 0. However, it definitely was much lower (way less than a US$1 million). A quick search on the Internet shows that around that time frame, there was a major avocado supply [shortage](https://laist.com/2017/08/17/avocado_shortage.php) in the US.