PSY6422 - Project
ID: 220251464
2023-05-16
STARBUCKS & COSTA COFFEE
Background & Research Question
Europe has been deemed to be the world’s largest coffee consumer (per
2017-2021), accounting for 32% of global consumption in 2021 (cbi.eu,
2022). The two biggest coffee shop chain brands in Europe are
Costa Coffee and Starbucks (Worldatlas.com/Economics,
2018). Therefore, many may be interested in the investigation of
core differences of Costa versus Starbucks coffee.
For this reason, I wanted to explore the caffeine content and size
difference in a selection of ‘standard’ drinks. Demonstrating the
differences in what one would receive if they ordered the same drink
from each brand. For the purpose of narrowing down the field of interest
and ease of visualisation, I chose to focus on these two aspects.
The research Question becomes as follows: “How do the caffeine content and drink volume, for an equal size, differ between Costa Coffee and Starbucks for a selection of ‘standard’ coffee drinks in Europe and the UK?”
Data Origins
The raw data was obtained via the Starbucks and Costa websites, publicly accessible:
Starbucks. (2023, February). Core Beverage Nutritional and
Allergen Information (pdf). Starbucks Nutrition and Allergen
Information. link
Costa. (2023, May). Costa in-store products: allergen information
guide (pdf). Costa Nutrition and allergens info. link
Fourteen ‘core’ coffee drink types were selected, each with existing
as a Starbucks and Costa version. These are all on their general menus
and the selection allows for a non-crowded visualization of ‘core’ types
of coffee as well as equal comparison. Caffeine content (in milligrams
(mg)) and drink volume (in milliliters (mL)) of these coffee drinks were
inserted into an excel datasheet. This sheet makes up the raw
data.
The data comprises the reported values as of May 2023 for Europe and UK
in medium size (or standard if no medium exists) takeaway drinks.
Project Organization
The /data folder contains data required for the project, and /images contains images required for the project, as well as visualization outputs.
A codebook describing all labels and abbrevations used in this project for data, variables, functions etc. within this project is located at /codebook.xlsx.
Data Preparation
Loading packages
The project utilized the renv package to retain package versions, safeguarding it from potential updates to packages in the future.
Package versions used in this project are listed within the file /renv.lock
#Load packages with renv
install.packages("renv")
library(renv)
renv::restore()
#Import packages
library(tidyverse)
library(gapminder)
library(ggplot2)
library(png)
library(RCurl)
library(grid)
library(dplyr)
library(scales)
library(showtext)
library(here)
library(readxl)
library(knitr)
library(kableExtra)
library(readr)Importing data
#Specify image relative paths for logos and legend icons
Starbucks_image <- readPNG((here::here("images", "starbucks_logo.png")),
native = TRUE)
Costa_image <- readPNG((here::here("images", "costa_logo.png")),
native = TRUE)
starbucks_cup <- readPNG((here::here("images", "starbucks_cup.png")),
native = TRUE)
costa_cup <- readPNG((here::here("images", "costa_cup.png")),
native = TRUE) #Read raw data
rawdata <- read_excel(here::here("data","DAAVDATA.xlsx"))
kable(rawdata, format = "markdown")| Type of coffee | Volume (mL) Starbucks | Volume (mL) Costa | Caffeine (mg) Starbucks | Caffeine (mg) Costa |
|---|---|---|---|---|
| Decaf Coffee | 455 | 382 | 2 | 2 |
| Single-shot Espresso | 25 | 30 | 33 | 100 |
| Frappuccino / Frappé | 455 | 499 | 33 | 100 |
| Cold Brew | 455 | 345 | 50 | 210 |
| Cortado | 170 | 180 | 66 | 141 |
| Double-shot Espresso | 50 | 60 | 66 | 200 |
| Latte | 455 | 364 | 66 | 200 |
| Iced Latte | 455 | 473 | 66 | 200 |
| Flat White | 227 | 300 | 66 | 241 |
| Filter / Brewed Coffee | 455 | 382 | 136 | 256 |
| Mocha | 455 | 332 | 66 | 325 |
| Cappucino | 455 | 362 | 66 | 325 |
| Triple-shot Espresso | 75 | 90 | 99 | 325 |
| Americano | 455 | 340 | 99 | 325 |
Cleaning the data
#Create clean dataframe
cleandata <- data.frame(
Type.of.Coffee = c(rep(rawdata$'Type of coffee', 2)),
Brand = rep(c("Starbucks" , "Costa"), 14),
Caffeine = c(rawdata$`Caffeine (mg) Starbucks`,
rawdata$`Caffeine (mg) Costa`),
Volume = c(rawdata$`Volume (mL) Starbucks`, rawdata$`Volume (mL) Costa`))
#Define the desired orders
order_list <- c(1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10,
11, 11, 12, 12, 13, 13, 14, 14)
order_list2 <- c(1, 15, 2, 16, 3, 17, 4, 18, 5, 19, 6, 20, 7, 21, 8, 22, 9, 23,
10, 24, 11, 25, 12, 26, 13, 27, 14, 28)
#Define a function to reorder a vector based on an order list
reorder_vec <- function(vec, order_list) {
vec[order_list]}
#Reorder the columns using the reorder_vec function
cleandata <- cleandata %>%
mutate(Type.of.Coffee = reorder_vec(Type.of.Coffee, order_list),
Caffeine = reorder_vec(Caffeine, order_list2),
Volume = reorder_vec(Volume, order_list2)) #Specify order of Coffee
cleandata$Type.of.Coffee <- factor(cleandata$Type.of.Coffee,
levels = unique(cleandata$Type.of.Coffee),
ordered = TRUE)
#(see visualisation section for final dataframe)Visualisation
#Define data to plot
ggp <- ggplot(cleandata, aes(fill=Brand, y=Caffeine, x=Type.of.Coffee))
#Display clean and processed data
kable(cleandata, format = "markdown")| Type.of.Coffee | Brand | Caffeine | Volume |
|---|---|---|---|
| Decaf Coffee | Starbucks | 2 | 455 |
| Decaf Coffee | Costa | 2 | 382 |
| Single-shot Espresso | Starbucks | 33 | 25 |
| Single-shot Espresso | Costa | 100 | 30 |
| Frappuccino / Frappé | Starbucks | 33 | 455 |
| Frappuccino / Frappé | Costa | 100 | 499 |
| Cold Brew | Starbucks | 50 | 455 |
| Cold Brew | Costa | 210 | 345 |
| Cortado | Starbucks | 66 | 170 |
| Cortado | Costa | 141 | 180 |
| Double-shot Espresso | Starbucks | 66 | 50 |
| Double-shot Espresso | Costa | 200 | 60 |
| Latte | Starbucks | 66 | 455 |
| Latte | Costa | 200 | 364 |
| Iced Latte | Starbucks | 66 | 455 |
| Iced Latte | Costa | 200 | 473 |
| Flat White | Starbucks | 66 | 227 |
| Flat White | Costa | 241 | 300 |
| Filter / Brewed Coffee | Starbucks | 136 | 455 |
| Filter / Brewed Coffee | Costa | 256 | 382 |
| Mocha | Starbucks | 66 | 455 |
| Mocha | Costa | 325 | 332 |
| Cappucino | Starbucks | 66 | 455 |
| Cappucino | Costa | 325 | 362 |
| Triple-shot Espresso | Starbucks | 99 | 75 |
| Triple-shot Espresso | Costa | 325 | 90 |
| Americano | Starbucks | 99 | 455 |
| Americano | Costa | 325 | 340 |
Colours
Colours of bars in the visualisation correspond to iconic colours of
the brands, for the Caffeine content, increasing glance value for those
familiar with brands. This is important as those most likely to be
interested in the data are those familiar with the brands.
Additionally, for better clarity, the bars corresponding to the drink
Volume, are coloured coffee-like colours, a light contrast with the
Caffeine bars is ensured by slightly fading them and brands are
differentiated by the two different hues of brown.
(see Design section)
Custom Text
Using the showtext package, a personalized text was imported and incorporated into the visualization. The chosen font closely resembles the fonts employed by the two brands, thereby enhancing the professional appearance and overall attractiveness of the visualization.
font_add_google(name = "Source Sans Pro", family = "Source Sans Pro")
#Load new custom font for showtext package
showtext_auto()
#Automatically use showtext for plot Design
The logos and icons imported earlier are used in the legends for the
graph. They provide a visual appeal of immediate recognition for the
logos, which correspond to the bar colours, and the coffee cup icons
link in with the overall theme of the investigation: Coffee. These
coffee cups are coloured correspondingly with the bars which represent
each brands’ drink volume.
The brighter logo-colours where chosen to represent Caffeine to draw
more attention to this statistical representation, as compared to the
coffee-colours representing the drink volume, as seemed fitting for the
liquid measurement. The coffee-colours were made slightly transparent,
making them lighter to provide a better contrast,
The decision to use grouped barcharts with two y-axis was based on
the fact that this would best visualise the contrast between the two
brands for both the caffeine content and the drink size
difference. Especially, due to the fact that a small drink can still
have a lot of caffeine and vice versa. Decaf coffee was especially
included to reflect this and for the same reason, it was placed first on
the x-axis as a point of comparison.
The rest of the data was ordered on the basis of least to most caffeine
for both of the Brands, for each type of coffee. Meaning,
although Costa Cold Brew had a higher caffeine content than Costa
Cortado, Starbucks Cold Brew did not have a higher caffeine content than
Starbucks Cortado, so, Cold brew is put first on the x-axis. The volume
data is dependent on the type of drink for each brand and is therefore
ordered based on the caffeine data. Descriptive labels were included to
be able to gain an understanding from the visualisation if it were to be
presented on its own.
#Plot graph and customize various plot aesthetics
graph <- ggp +
#adding volume barplot:
geom_bar(aes(y = Volume / 1,
fill = ifelse(Brand == "Starbucks", "C", "D")), #seperate bar colours
position = "dodge", stat="identity", #specifying a grouped barplot
width = 0.8, alpha=.5, #altering bar width and transparancy,
show.legend = FALSE) + #hiding default legend
#adding caffeine barplot, specifying colours
geom_bar(aes(fill = ifelse(Brand == "Starbucks", "A", "B")),
position = "dodge", stat = "identity", #making it a grouped barplot
width = 0.5, #narrowing bar width to better see plot bars
show.legend = FALSE) + #hiding default legend
#naming left y-axis, changing axis breaks, making space for legends
#and removing gap between graph and axis:
scale_y_continuous(name = "Caffeine (mg)",
breaks = seq(0, 500, by = 50),
limits = c(0, 670), expand =c(0,0),
#adding secondary y-axis:
sec.axis = sec_axis(~.*1, name="Volume (mL)",
breaks = seq(0, 500, by = 50))) + #specyfing axis breaks
ggtitle("Costa VS Starbucks: Caffeine & Size") + #creating the title
labs(x = "Type of Coffee*", #labelling the x-axis
subtitle = "Europe & UK, 2023", #creating the subtitle
caption = "*Takeaway, Standard/Medium Size") + #and a x-axis caption
#specifying colours for the different bars:
scale_fill_manual(values = c("A" = c("#1b703f"), "B" = c("#B91345"),
"C" = c("#63330b"), "D" = c("#996633"))) +
theme(aspect.ratio = 3/5, #setting the proportions of the plot
panel.grid.major = element_blank(), #removing the plot grid
panel.grid.minor = element_blank(),
panel.background = element_blank(), #removing the plot background
plot.margin = unit(c(1,1,2,1), "cm"), #changing plot margin around it
#changing colour + thickness of axis lines:
axis.line = element_line(colour = "black", linewidth = 1),
#adjusting x-axis break labels: direction, alignment and axis distance:
axis.text.x = element_text(angle = 50, vjust = 0.5,
hjust = 1, margin = margin(t = -30)),
#adjusting x-axis label title: distance and text size:
axis.title.x = element_text(margin = margin(t = 70), size = 12),
#adjusting left y-axis label: distance:
axis.title.y = element_text(margin = margin(r = 15)),
#adjusting right y-axis label: distance:
axis.title.y.right = element_text(margin = unit
(c(0, 0, 0, 5), 'mm')),
#altering plot title aesthetics:
plot.title = element_text(margin = margin(b = 6),
hjust = 0.5, size = 14, face = "bold"),
#changing subtitle aesthetics:
plot.subtitle = element_text(margin = margin (b = 20),
size = 11, hjust = 0.5),
#altering the captions aesthetics:
plot.caption = element_text(margin = margin (t = 10),
color = "#6e6e6e", face = "italic",
hjust = 0.5)) +
#
#creating my own legend:
#adding Starbucks logo + coordinates:
annotation_raster(Starbucks_image, xmin = 0.73, xmax = 1.39,
ymin = 521, ymax = 573) +
#adding Costa logo + coordinates:
annotation_raster(Costa_image, xmin = 0.491, xmax = 1.63,
ymin = 578, ymax = 630) +
#adding cup icon for Starbucks Volume + coordinates:
annotation_raster(starbucks_cup, xmin = 13.5, xmax = 14.5,
ymin = 515, ymax = 570) +
#adding cup icon for Costa Volume + coordinates:
annotation_raster(costa_cup, xmin = 13.5, xmax = 14.5,
ymin = 575, ymax = 630) +
#adding legend text on the left:
annotate("text", x = 1.7:1.7, y = c(550, 605),
label = c("Starbucks", "Costa"), hjust = 0) +
#adding legend text on the right:
annotate("text", x = 13.5:13.5, y = c(543, 603),
label = c("Starbucks", "Costa"), hjust = 1) +
#adding legend title on the left:
annotate("text", x = 0.8, y = 659, label = "Brand",
hjust = 0, fontface = "bold") +
#adding legend title on the right:
annotate("text", x = 14.2, y = 658, label = "Brand Volume",
hjust = 1, fontface = "bold")Saving Visualisation
#Save plot to images folder
windows(width = 1000, height = 800) #Open windows graphics device
print(graph)
dev.print(file = here("images", "viz220251464.png"), device = png,
width = 1000, height = 800)Result
Interpretation & Future Direction
It appears that Costa coffees have more caffeine in them in
comparison to Starbucks coffees, as this was shown for all drink
types.
Additionally, there does not seem to be a big difference in drink size
(volume) when contrasting the two brands. However, Starbucks does seem
to have slightly larger drinks across the majority of the types of
coffees.
Thus, although you may get slightly more coffee in your drink at
Starbucks it seemingly has less than half the amount of caffeine as
compared to a Costa coffee overall.
Furthermore, the graph informs on the relative differences between the
coffees, such as the indication that there is just as much caffeine in
an Americano as there is in a Triple-Shot Espresso.
Further aspects to investigate could be price differences as well as other nutritional information such as sugar content. This would best suit additional visualisations, for clarity, as for example sugar in grams would be quite alot higher than Caffeine milligrams and would not accurately demonstrate the individual differences.
Summary
I have improved my knowledge of R and utilization of R functions, R
Markdown, and Github. The script serves as a testament to my newfound
understanding of employing R methods for data manipulation, importing,
cleaning, and visualization.
Additionally, I have improved the reproducibility of my work by using
Renv for an automated package management system.
Moreover, the project showcases my proficiency in creating customized
visualizations featuring multiple bar plots, axes, legends, and
annotations, all aimed at effectively conveying information. As a result
of this experience, I have also gained the ability to leverage R
Markdown to produce dynamic reports and seamlessly integrate code,
visualizations, and text.