Twitter #SWDcommunity challenge https://twitter.com/storywithdata/status/1274019877779619841 – https://twitter.com/ikashnitsky/status/1274092680407113730

improve-plot.r

#===============================================================================
# 2020-06-19 -- twitter
# improve plot
# Ilya Kashnitsky, ilya.kashnitsky@gmail.com
#===============================================================================

# the challenge
# https://community.storytellingwithdata.com/exercises/one-little-changeand-a-redesign

library(tidyverse)
library(magrittr)
library(ggdark)
library(hrbrthemes)
library(lubridate)

# data
raw <- tibble::tribble(
    ~BEACON, ~DRAPER, ~FILMORE, ~LAKESIDE, ~MARE.VALLEY, ~NORTH, ~OAKLEY, ~ORLY, ~PIERCE, ~ROSEDALE, ~SEALY, ~SOUTHLAKE, ~WESTLAKE, ~WILDLAND, ~REGIONAL.AVG, 
    69L, 130L, 85L, 117L, 16L, 74L, 61L, 51L, 35L, 65L, 93L, 72L, 83L, 83L, 74L, 
    77L, 86L, 97L, 107L, 17L, 71L, 55L, 61L, 32L, 55L, 89L, 64L, 73L, 79L, 69L, 
    63L, 90L, 92L, 127L, 18L, 68L, 55L, 67L, 32L, 59L, 79L, 58L, 73L, 81L, 69L, 
    61L, 100L, 95L, 132L, 21L, 72L, 72L, 51L, 49L, 53L, 82L, 77L, 85L, 75L, 73L, 
    67L, 106L, 92L, 145L, 21L, 69L, 79L, 51L, 49L, 59L, 94L, 87L, 87L, 79L, 78L, 
    67L, 91L, 83L, 112L, 17L, 71L, 60L, 66L, 42L, 52L, 83L, 70L, 54L, 72L, 67L, 
    57L, 89L, 103L, 107L, 17L, 56L, 42L, 52L, 23L, 41L, 84L, 66L, 59L, 51L, 61L, 
    62L, 87L, 109L, 95L, 17L, 66L, 44L, 60L, 48L, 37L, 81L, 59L, 64L, 46L, 63L, 
    68L, 93L, 94L, 126L, 21L, 83L, 65L, 66L, 38L, 63L, 89L, 66L, 59L, 66L, 71L, 
    61L, 96L, 104L, 133L, 13L, 79L, 80L, 47L, 24L, 49L, 70L, 65L, 68L, 57L, 68L, 
    54L, 103L, 88L, 105L, 22L, 66L, 37L, 63L, 26L, 52L, 66L, 55L, 51L, 69L, 61L
) %>% 
    mutate(quart = c("2017 Q1", "2017 Q2", "2017 Q3", "2017 Q4", "2018 Q1", "2018 Q2", "2018 Q3", "2018 Q4", "2019 Q1", "2019 Q2", "2019 Q3")
)

# separate out the regional average
df_reg <- raw %>% transmute(quart, reg_avg = REGIONAL.AVG)

# cleen the rest of the data
df <- raw %>% 
    select(-REGIONAL.AVG) %>% 
    pivot_longer(BEACON:WILDLAND, names_to = "dealer") %>% 
    # orger dealers by average sales
    group_by(dealer) %>% 
    mutate(avg_sales = value %>% mean) %>% 
    ungroup() %>% 
    mutate(
        dealer = dealer %>% 
            str_to_title() %>% 
            str_replace(".v", " v") %>% 
            as_factor() %>% 
            fct_reorder(avg_sales) %>% 
            fct_rev()
    )

# minimal labels
quart_lab <- c("2017", "", "", "", "18", "", "", "", "19", "", "")

# plot
df %>% 
    mutate(new = dealer) %>%  
    ggplot(aes(quart, value))+
    geom_hline(yintercept = 0)+
    geom_path(
        data = . %>% select(-dealer),
        aes(group = new),
        # color = "#ababab",
        size = .5,
        alpha = .25
    )+
    geom_path(
        data = df_reg, aes(y = reg_avg, group = 1),
        size = .5, color = "gold"
    )+
    geom_path(aes(group = dealer), size = 1)+
    facet_wrap(~dealer, ncol = 5)+
    scale_x_discrete(labels = quart_lab, expand = c(0,0))+
    scale_y_continuous(sec.axis = dup_axis())+
    labs(title = "Cars sold by dealership per quarter",
         caption = "@ikashnitsky",
         x = NULL, y = NULL)+
    geom_text(
        data = . %>% filter(dealer == "Mare valley", quart == "2019 Q3") %>%
            transmute(dealer, quart, value = 90),
        label = "Regional average", hjust = 1, size = 3,
        color = "gold", family = font_rc
    )+
    dark_theme_minimal(base_family = font_rc)+
    theme(
        plot.title = element_text(family = "Roboto Slab", size = 20),
        panel.spacing.x =  unit(1, "lines"), 
        panel.grid.minor = element_blank(),
        panel.grid.major.x = element_line(
            size = .25, color = "#cdcdcd33"
        )
    )

# save the result
ggsave("improved.png", width = 8, height = 5)