How To: Visualize Big Tech Stocks the Tidy Tuesday Way
A tutorial to go along with my 2023 Week 6 submission for Tidy Tuesday
For this week’s Tidy Tuesday1 I created the following chart in R and posted it on Twitter.
Below I'll walk through a step-by-step guide on how to create this combined small multiples line chart and stream graph using R2. The goal was to create a graphic that had the feeling of something you would see in a newspaper layout such as in the Financial Times or Wall Street Journal, two of the most data visualization-savvy and finance-focused publications out there.
Getting Started
This tutorial assumes you've installed R and have a basic understanding of the tidyverse packages such as {dplyr} and {ggplot2}. But nothing more than that is required.
First, let's install the core packages needed for the visualization we're going to create.
# load packages
library(tidyverse)
library(ggtext)
library(ggimage)
library(ggstream)
library(scales)
library(patchwork)
library(gganimate)
Now that we've installed the right packages I'm going to set our custom plot theme so that our visuals have a minimalist look to them with an off-white background. The inspiration comes from Owen Phillips' tutorials in his now defunct newsletter The F5.
# Custom ggplot theme (inspired by Owen Phillips at the F5 substack blog)
theme_custom <- function () {
theme_minimal(base_size=11, base_family="Outfit") %+replace%
theme(
panel.grid.minor = element_blank(),
plot.background = element_rect(fill = 'floralwhite', color = "floralwhite")
)
}Next, we read in the data from the Tidy Tuesday Github repository and pre-process by assigning colors for each company and creating a few variables that will help us out in later steps.
## --------------- Data Wrangling ---------------------------
big_tech_companies <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2023/2023-02-07/big_tech_companies.csv")
# Assign companies colors
big_tech_companies$clr <- c("#A2AAAD", "#FF0000", "#FF9900", "#1798c1", "#00BCEB", "#34A853", "#0530AD", "#0068B5", "#0668E1", "#F14F21", "#E50914", "#76B900", "#C74634", "#E31937")
# Companies logos (downloades in folder Logos)
big_tech_companies <- big_tech_companies |>
mutate(
lower_stock = tolower(stock_symbol),
company_logo_path = paste0(here::here("Logos/"), lower_stock, ".png")
)
big_tech_stock_prices <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2023/2023-02-07/big_tech_stock_prices.csv")
big_tech_stock_prices <- big_tech_stock_prices |>
left_join(big_tech_companies) |>
mutate(dollar_volume = adj_close*volume) |>
mutate(duplicate = stock_symbol)
# Get company max closing value and min date
companies_logos_df <- big_tech_stock_prices |>
group_by(stock_symbol, company_logo_path) |>
summarise(
first_date = min(date),
max_adj_close = max(adj_close)
) |>
ungroup()The resulting data should look something like this:
Stock Price Facet
From there we can create our first plot.
## 1. ----------------- Big Tech Facet Plot ---------------------
# Main plot: company facet
big_tech_plot <- ggplot() +
geom_line(data = big_tech_stock_prices, aes(date, adj_close, color = clr, group = stock_symbol), size = .75) +
#geom_line(data = big_tech_stock_prices, aes(date, adj_close, color = 'grey', alpha = 0.5, group = duplicate), size = .75) +
geom_ribbon(data = big_tech_stock_prices, aes(date, ymin = 0, ymax = adj_close, fill = clr), alpha = .6) +
geom_richtext(
data = companies_logos_df, aes(x = first_date, y = max_adj_close, label = glue::glue("<img src='{company_logo_path}' width='35'/>")),
hjust = 0,
fill = NA,
label.padding = unit(.125, "lines"),
label.r = unit(0, "lines"),
label.size = unit(0, "lines")
) +
labs(
title = "Big Tech has Gotten Bigger",
subtitle = "Evolution of select technology stock prices between 2010 and 2022",
caption = "Data: Yahoo Finance/Evan Gower | Graphic: @steodosescu"
) +
scale_color_identity() +
scale_fill_identity() +
coord_cartesian(clip = "off", expand = F) +
facet_wrap(vars(stock_symbol), scales = "free", nrow = 5) +
#facet_wrap(~fct_reorder(stock_symbol, dollar_volume), scales = "free", nrow = 5) +
theme_custom() +
theme(
plot.title = element_text(color = "black", face = "bold", family = "Outfit", hjust = 0.5, size = rel(3.0), margin = margin(b = .5, unit = "cm")),
plot.subtitle = element_text(color = "#444444", family = "Outfit", hjust = 0.5, size = rel(1.2), margin = margin(b = 1, unit = "cm")),
strip.text = element_blank(),
panel.spacing.x = unit(1, "cm"),
panel.spacing.y = unit(1.5, "cm"),
panel.grid = element_blank(),
axis.title = element_blank(),
axis.line.x = element_line(),
axis.text.x = element_text(color = "black", family = "Outfit", size = rel(1.0)),
axis.text.y = element_blank(),
axis.ticks.length = unit(.125, "cm"),
axis.ticks.x = element_line(size = .25)
)
big_tech_plot
Stream Chart
The second plot is a stream chart showing the dollar volumes (closing price each day * volume of daily shares being traded) of six of the biggest US tech companies over time.
To create the graphic we first isolate the colors of the select companies and then use geom_stream() from the {ggstream} package to get our stream plot3.
# isolate colors
big_six_colors <- big_tech_companies %>%
filter(lower_stock %in% c('aapl', 'msft', 'googl', 'amzn', 'nvda', 'nflx'))
# make stream plot
dollar_volume_plot <- big_tech_stock_prices %>%
filter(lower_stock %in% c('aapl', 'msft', 'googl', 'amzn', 'nvda', 'nflx')) %>%
ggplot(aes(x = date, y = dollar_volume, fill = stock_symbol)) +
geom_stream(type = 'ridge', color = 'black', alpha = .95, n_grid = 4900) + # n_grid needs to be at least 2% more than the number of observations per group
scale_fill_manual(values = big_six_colors$clr) +
#scale_y_continuous(labels = scales::dollar())+
scale_y_continuous(labels = label_number(accuracy=0.1, suffix = "B", scale = 1e-9)) +
scale_x_continuous(breaks = seq(2010, 2022, 1)) +
coord_cartesian(clip = 'off') +
theme_custom() +
theme(legend.position = "bottom",
legend.title = element_blank(),
axis.text.x = element_text(size = 10,
vjust = 2.75,
hjust = 1.5),
panel.grid.major.x = element_blank(),
plot.title.position = 'plot',
plot.title = element_text(face = 'bold',
hjust = 0.5,
size = 20),
plot.subtitle = element_text(hjust = 0.5),
plot.margin = margin(10, 10, 15, 10)) +
labs(x = "",
y = "Close Price * Volume ($)",
title = "Big Tech Trading Volumes",
subtitle = "Dollar volumes of six of the US's biggest tech stocks have increased overall, but are in decline recently.",
caption = "Data: Yahoo Finance/Evan Gower | Graphic: @steodosescu")
dollar_volume_plot
The last thing is to combine the two plots. We do this using the {patchwork} package by using simple syntax that stitches one graphic on top of the other with the following command: big_tech_plot / dollar_volume_plot. Finally, we add some theming to ensure everything looks clean and save it in a size that is appropriate before being shared.
## Combine plots with patchwork
big_tech_plot / dollar_volume_plot +
theme(
panel.background = element_rect(fill= "floralwhite", color=NA),
plot.background = element_rect(color = "floralwhite", fill = "floralwhite"),
plot.margin = margin(1, 0.5, 0.5, 0.5, unit = "line")
)
ggsave("2023/W6_Big Tech Patchwork.png", device = ragg::agg_png, dpi = 300, width = 8, height = 16)The end result:
Bonus: Stock Price Animation
After I published my initial graphic seen above I came across a great tweet from Ansgar Wolsing showing what a $100 investment in one of the big tech stocks 10 years ago would net out to every month from January 2013 to today using this same data.
I decided to recreate the graphic using my personal theming elements. Let’s walk through how to do it.
First, we filter for 10 years of data and create a simple static bar chart that replaces the company labels with their logos on the y axis. From there we calculate the monthly adjusted close prices at the beginning and end of each month, join in the company colors and store in a dataframe called df_plot_anim.
Then we create the ggplot object p that adds in all the final theming elements. At this point the plot looks like a jumble of logos and bars because we are asking ggplot to plot each stock in rank order for every single month for the past 10 years. The next step will be to animate the graphic using the close prices for each of those months so that the bars appear to be moving fluidly over time.
## 3. -------- Animated stock prices -----------------
# inspiration from Ansgar Wolsing: https://github.com/bydata/tidytuesday/blob/main/2023/06/eda.R
start_date <- as_date("2013-01-01")
max_date <- as_date("2022-12-29")
pal <- c("#A2AAAD", "#FF0000", "#FF9900", "#1798c1", "#00BCEB", "#34A853",
"#0530AD", "#0068B5", "#0668E1", "#F14F21", "#E50914", "#76B900", "#C74634", "#E31937")
df_plot <- big_tech_stock_prices %>%
filter(date >= start_date) %>%
group_by(stock_symbol) %>%
mutate(adj_close_rel = 100 * adj_close / .$adj_close[.$stock_symbol == stock_symbol & .$date == min(.$date)]) %>%
# filter(date == max_date) %>%
ungroup() %>%
# quick & dirty formatting of company names
mutate(company2 = str_extract(company, "[^\\s,\\.]+"),
company2 = ifelse(stock_symbol == "IBM", "IBM", company2),
company2 = fct_reorder(company2, adj_close_rel),
company_label = glue::glue("<img src='{company_logo_path}' width='30'/>")) %>%
select(stock_symbol, company = company2, company_logo_path, company_label, date, adj_close_rel)
# test it
df_plot %>%
filter(date == max_date) %>%
ggplot(aes(company, adj_close_rel)) +
geom_col(aes(fill = big_tech_companies$clr)) +
geom_richtext(aes(label = company_label)) +
scale_fill_identity()+
coord_flip()
bar_offset <- 0.35
df_plot_anim <- df_plot %>%
# filter(date >= as_date("2018-01-01")) %>%
filter(date <= max_date) %>%
mutate(month = floor_date(date, "1 month")) %>%
arrange(date) %>%
# calculate summary statistic per month
group_by(company, company_label, month) %>%
summarize(adj_close_rel_first = first(adj_close_rel),
adj_close_rel_last = last(adj_close_rel),
adj_close_rel_avg = mean(adj_close_rel), .groups = "drop") %>%
group_by(month) %>%
mutate(rank = rank(-adj_close_rel_first), ties.method = "first") %>%
ungroup()
# join in colors
big_tech_companies2 <- big_tech_companies %>%
mutate(company2 = str_extract(company, "[^\\s,\\.]+"),
company2 = ifelse(stock_symbol == "IBM", "IBM", company2))
df_plot_anim <- df_plot_anim %>%
left_join(big_tech_companies2, by = c("company" = "company2"))
p <- df_plot_anim %>%
# filter(month == max(month)) %>%
ggplot(aes(adj_close_rel_last, rank)) +
# use geom_rect instead of geom_col for smooth transitions of the bars
geom_rect(
aes(ymin = rank - bar_offset, ymax = rank + bar_offset,
xmin = 0, xmax = adj_close_rel_last,
fill = clr)
) +
# company icons
geom_richtext(
aes(x = -6, label = company_label, hjust = 1),
label.size = 0, fill = NA
) +
# stock value
geom_text(
aes(label = paste("$", round(adj_close_rel_last))),
size = 3, family = "Outfit", hjust = -0.15) +
scale_x_continuous() +
scale_y_reverse() +
scale_fill_identity() +
coord_cartesian(clip = "off") +
#guides(fill = "none", color = "none") +
labs(
title = "What's it Worth Today?",
subtitle = "...if you had invested $100 in one of these Tech stocks 10 years ago
<br>
<br><span style='font-size: 16pt; color: #888888; font-family: Outfit'>
{format(frame_time, '%Y (%B)')}</span>",
x = "Close Price",
y = "",
caption = "Closing price of the first day of a month after adjustments
for all applicable splits and dividends shown. |
**Source:** Yahoo Finance |
**Graphic:** @steodosescu"
) +
theme_custom() +
theme(
plot.margin = margin(t = 2, b = 2, l = 18, r = 18),
plot.title = element_text(size = 20, family = "Outfit", face = "bold", hjust = 0.5),
plot.title.position = "plot",
plot.subtitle = element_markdown(lineheight = 1.1),
plot.caption = element_textbox(size = 7, width = 1, lineheight = 1.1,
margin = margin(t = 4, b = 8))
)
To animate the bar chart we use the {gganimate}package. Using the function transition_time() we specify on what timeframe the frames should change (month in this case), and then call the appropriately names animate() function which does the heavy lifting in making the graphic come to life.
# Animate
p_anim <- p +
transition_time(month) +
view_follow(fixed_x = FALSE, fixed_y = TRUE)
# this part takes a while
fps <- 12
animate(p_anim, ref_frame = 1, nframes = length(unique(df_plot_anim$month)) * 5,
end_pause = 2 * fps,
res = 150, width = 800, height = 800, fps = fps)
anim_save("hindsight-bias.gif")Conclusion
Today we learned how to create small multiples graph, a stream chart and an animated bar chart that shows how your investments in Big Tech may have done over time.
If you enjoyed this tutorial feel free to subscribe to the free newsletter below or reach out on Twitter. If you found this useful in your data visualization journey with R please share the post, or if so inclined, Buy me a Coffee at the link below.
If you want more content like this feel free to subscribe to the newsletter as well.
Tidy Tuesday is a weekly data project spearheaded by Tom Mock at Rstudio. Its intent is to provide a public space for people to share data wrangling and visualization skills using a new dataset each week. It’s primarily aimed at users of the R programing language ecosystem, which many submitters use to generate interesting plots and post on Twitter for the community to see.
One thing I learned as a part of this is the n_grid parameter within geom_stream() needs to be a value at least two percent greater than the number of observations in your dataset. This attribute controls the smoothness of the stream lines.