hotels script.Rmd

---
title: "Hotel Bookings"
author: "EE"
date: "2/13/2020"
output: html_document
---

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

library(tidyverse)
library(gghighlight)
library(ggtext)
library(sysfonts)
library(showtext)
library(countrycode)
library(lubridate)

set.seed(0408)


hotels <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2020/2020-02-11/hotels.csv')

countries <- codelist_panel %>%
  select(country.name.en, iso3c, continent)

background_col <- "#251e3e"
text_col <- "grey90"
min_var <- "#a3307e"
max_var <- "#fea873"

font_add_google("Lobster", "Lobster")

showtext_auto()
```

```{r data manipulation}
euro_hotels_ranked <- hotels %>%
  mutate(quarter = case_when(
    is.element(arrival_date_month, c("January", "February", "March")) ~ "Q1",
    is.element(arrival_date_month, c("April", "May", "June")) ~ "Q2",
    is.element(arrival_date_month, c("July", "August", "September")) ~ "Q3",
    is.element(arrival_date_month, c("October", "November", "December")) ~ "Q4"
  ),
        yr_qtr = paste0(arrival_date_year, ":", quarter) %>%
          yq()) %>%
  count(country, arrival_date_year, quarter, yr_qtr, sort = TRUE) %>%
  left_join(countries, by = c("country" = "iso3c")) %>%
  filter(continent == "Europe") %>%
  distinct(country, yr_qtr, n, .keep_all = TRUE) %>%
  add_count(country, name = "cnt") %>%
  filter(cnt == 9) %>%
  group_by(yr_qtr) %>%
  mutate(rank = row_number(desc(n))) %>%
  ungroup() %>%
    group_by(country) %>%
  mutate(sd_rank = sd(rank)) %>%
  ungroup() %>%
  mutate(line_col = case_when(
           is.element(sd_rank, max(sd_rank)) ~ max_var,
           is.element(sd_rank, min(sd_rank)) ~ min_var,
           TRUE ~ "grey90"
         ))
  
```


```{r plot}

col_tbl <- euro_hotels_ranked %>%
  distinct(country.name.en, line_col)

cols <- col_tbl$line_col
names(cols) <- col_tbl$country.name.en


hotel_plot <- euro_hotels_ranked %>%
  ggplot(aes(x = yr_qtr, y = rank, color = country.name.en)) +
  geom_point(size = 2, alpha = .8) +
  geom_line(size = 1.5) +
  geom_text(data = euro_hotels_ranked %>%
                    filter(yr_qtr == "2015-07-01"), aes(x = yr_qtr - 10, label = country.name.en, color = country.name.en),
            hjust = 1,
            family = "Lobster",
            size = 14) +
  geom_text(data = euro_hotels_ranked %>%
                    filter(yr_qtr == "2017-07-01"), aes(x = yr_qtr + 10, label = country.name.en, color = country.name.en),
            hjust = 0,
            family = "Lobster",
            size = 14) +
  scale_y_reverse() +
  scale_x_date(
    breaks = as.Date(c("2015-01-01", "2016-01-01", "2017-01-01")),
    labels = c("", "2016", "2017")
  ) +
  scale_color_manual(
    values = cols
  ) +
  labs(
    x = "",
    y = "Rank",
    title = "Ranking European Countries by Number of Hotel Reservations",
    subtitle = "For each fiscal quarter between July 2015 and July 2017, <span style='color:#a3307e'>Portugal</span> had more hotel reservations than any other country. Due to a spike in reservations in Q2 2016, <span style='color:#fea873'>Serbia</span> shows the most variability in its quarterly rankings.",
    caption = "Data: Antonio, Almeida, & Nunes, 2019 | Viz: Eric Ekholm (@ekholm_e)"
  ) +
  expand_limits(x = as.Date(c("2015-01-01", "2017-12-31"))) +
  gghighlight(is.element(sd_rank, range(sd_rank)), use_direct_label = FALSE,
              unhighlighted_params = list(color = text_col)) +
  theme(
    rect = element_rect(fill = background_col),
    text = element_text(color = text_col, family = "Lobster", size = 24),
    plot.title = element_markdown(size = 40),
    plot.subtitle = element_textbox(size = 32, lineheight = 0),
    panel.background = element_rect(fill = background_col),
    plot.background = element_rect(fill = background_col),
    panel.grid = element_blank(),
    axis.text.x = element_text(color = text_col, size = 30),
    axis.text.y = element_blank(),
    axis.title.y = element_blank(),
    axis.ticks = element_blank(),
    legend.position = "none",
    plot.caption = element_markdown(size = 20)
  )

x11()
print(hotel_plot)

ggsave(here::here("2020 - 7 - hotel bookings/hotel plot.jpg"), device = "jpeg", width = 7*1.5, height = 7)
```