VizQuiz_5

The original graph(s): http://www.princeton.edu/~ina/infographics/starbucks.html

The following dashboard can be found in Princeton’s International Networks Archive, designed by ‘Flaming Toast Productions’ in far 2003.

And… the idea behind it is undoubtly good: it is bright, is it telling, it may be very attractive for some… color lovers (maybe, I’m not sure). But at the same time the charts there have some mistakes, which make the whole picture unreasonably heavy.

• 3d graphics - why? Was it fashionable those days? Look at the bar(?)chart at the lower left corner - what does it mean? Why 3d-cilinders, not 2d-bars? Why does the number of cilinders increasing to the 200… Oh, not, it does not increasing, just the volume of the bars increase… Okay, maybe this 3d vizualizaion is aimed to give more bright space to the right part of the chart and show the dynamics by this move, but… I do not think, that that is necessary. Moreover, why does the continious variables are shown in a barchart at all??!

Let’s have a look on the similar information, given by a 2d linechart:

library(datapasta)
library(dplyr)
#vector_paste()

year <- c(1987, 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003) %>% as.character() %>% as.numeric()
stores_n <- c(100, 200, 250, 300, 400, 500, 500, 500, 650, 900, 2400, 2500, 2850, 3100, 4200, 5500, 6200)%>% as.character() %>% as.numeric()

data_bar <- data.frame(year, stores_n)
library(knitr)
#kable(data_bar)
#str(data_bar)
#summary(data_bar)
library(ggplot2)

ggplot(data = data_bar, aes(x = year, y = stores_n)) +
  geom_line(linetype = "dashed", color="darkgrey") +
  geom_point(color="darkgreen", size=3) + 
  geom_label(
    data=data_bar %>% filter(stores_n > 3000),
    aes(label=stores_n), 
    nudge_x = 0.9, nudge_y = 0.9, 
    check_overlap = T) +
  labs(
    title = "Number of Starbacks stores worldwide", 
    subtitle = "6200 stores in total, with three new stores opening daily",
    caption = "Source: Starbacks.com, US Department of Agriculture, Fortune Magazine, 2003",
    x = "Year",
    y = "Number of stores") + 
  theme_bw()

In my honest opinion it shows the dynamics in a much clearer form, without unnecessary volume:

• Green color still responds to the brand color of Starbucks and grey dashline shows the angle of the growth;
• grey grid lines on a background helps to check the N of stores and years;
• labels at the higher point (after 2000 year) show an interesting jump in the number of opened stores;
• I have also changes a bit titling of the chart, without any text ON the actual chart as it was in the original, as well as put source under the graph and out a new title, placing an old one (with an idea) as a subtitle.

As I have understood, this dashboard is mainly focused on comparison of two firms. So, basically, if it needed, this chart can compare the data about these firms and give even more interersting ingormation. However, I do not have these info and I will not create it by myself. Just thinking out loud. But this chart could be used as a basis:

# для сравнения динамики открытия двух сетей можно использовать такую штуку: 
library(ggpubr)
ggscatterhist(
  data_bar, x = "year", y = "stores_n",
  margin.plot = "boxplot",
  ggtheme = theme_bw(), 
  title = "Number of Starbacks stores worldwide", 
  subtitle = "6200 stores in total, with three new stores opening daily",
  caption = "Source: Starbacks.com, US Department of Agriculture, Fortune Magazine, 2003",
  xlab = "Year",
  ylab = "Number of stores" 
  #, legend.title = "Stores:"
  )

The next vivid mistake is shown by the bottom left part:

• Why logos? Why logos of different sizes? It is extremly misleading. The reason is that authors did change not only the hight of each logo according to the y-axis, but they have enlarges is fully, which means that horizontal lenght on x-axis of every next logo becomes

Okay, so let’s try to recreate in a more simple view:bigger, too. So, basically, instead of seeing that McDonalds’ sales are 10 times bigger than Starbucks’ we see that McDonalds’ logo is 50 times bigger than Starbucks’! How I’ve calculated it? School math.

data_bar_2 <- data.frame(
stores = c("McDonalds", "AF's GDP", "BurgerKing", "Wendy's", "KFC", "PizzaHut", "TaсoBell", "Starbucks"),
sales = c(41, 21, 11.3, 9.4, 8.2, 8, 4.3, 4.1))
#kable(data_bar_2)
#str(data_bar_2)
#summary(data_bar_2)

ggplot(data = data_bar_2, 
       aes(x = reorder(stores, sales), 
           y = sales,
           fill=factor(ifelse(stores=="AF's GDP","Highlighted","Normal")))) +
  geom_bar(stat="identity") +
  scale_color_manual(name = "stores", values=c("red","grey50")) +
  geom_label(
    label=data_bar_2$sales,
    nudge_x = 0, nudge_y = 2, 
    check_overlap = T, 
    fill = "white"
  ) +
  labs(
    title = "Sales of stores worldwide", 
    subtitle = "Compared to GDP of Afghanistan",
    y = "Sales, billion",
    x = "") + 
  theme_bw() + theme(legend.position = "none")

So, here is…:

• no logos (such a pity). With a big wish we can Photoshop it. Also, we can change the colors of the bars, but I did not want to overload the graph. Ideally, the aim should be chosen first: what do want to show: comparison with Afghanistan’s GDP, comparison between McDonalds and Starbucks, or sales of each store in general? I have chosen the first option, which explains my color choice;
• labels with sales now are places directly on each bar, so the viewers do not need to play a labirinth game to connect label with a bar;
• I would also change the order of the bars in descending direction, but okay, if the client wants this option - I can handle that.

As for the maps… Well. the circles are misleading again - if it is placed on one dashboard I would make it the same sizes on both charts. Maybe that would tell the story in a clearer way. But in general - the things, that I have wanted to change mostly, are now changes.

Hope you enjoy this project as much as I have! Have a nice day and have more nice charts!