R Notebook

library(highcharter)

d1 = read_csv("tech_stocks_csv.zip", show_col_types = FALSE) %>% 
  filter(sale > 0) %>% 
  mutate(conm = stringr::str_to_title(conm), # Converts the string to title case
         datadate = lubridate::ymd(datadate)) # Convert datadate into a date variable

I will also set the Highcharter theme for all the plots

options(highcharter.theme = hc_theme_darkunica())

Q1

Note that while creating d1_1, I arranged it in the end so that we will get ordered bar plot.

d1_1 = d1 %>%
  group_by(conm) %>% 
  summarize(avg_sale = mean(sale), .groups = "drop") %>% 
  arrange(-avg_sale)

hchart(d1_1, "column", hcaes(x = conm, y = round(avg_sale, 2))) %>%
  hc_xAxis(title = list(text = "Company")) %>%
  hc_yAxis(title = list(text = "Average Sale in $ Millions"), 
           labels = list(format = "${value:,.0f}")) %>% # I used ChatGPT to find out how to create this format
  hc_chart(inverted = TRUE)

Q3

Here I will first create a scatterplot. Adding a regression line is a little more complex so I will add that below.

Also, I am rounding the numbers in the calculated ratios.

d1_ratio = d1 %>% 
  mutate(debt_ratio = round((dlc + replace_na(dltt, 0)) / at, 3),
         rnd_int = round(xrd / sale, 3))

d1_ratio %>% 
hchart("scatter", hcaes(x = debt_ratio, y = rnd_int)) %>% 
  hc_xAxis(title = list(text = "Debt Ratio")) %>%
  hc_yAxis(title = list(text = "R&D to Sales Ratio"))

Now let’s add a regression line. There is no direct way to add a regression line easily. We must estimate the regression model first and then use that model to get predicted values. I am using a super useful package in R called broom which is a part of tidyverse to convert the regression model into a data frame of predicted values.

library(broom)
reg = lm(rnd_int ~ debt_ratio, data = d1_ratio)
reg_fit = augment(reg) %>% arrange(debt_ratio)

augment function from broom package adds predicted values from the regression and other statistics back to the original data frame, which is d1_raio in our case. Take a peek at the resulting data frame:

head(reg_fit)

Here, we are intereted in the variable called .fitted which has predicted values of rnd_int based on debt_ratio. I arranged the data frame by debt_ratio so that highcharter can simply superimpose the regression line on our scatterplot.

d1_ratio %>% 
  hchart(type = "scatter", hcaes(x = debt_ratio, y = rnd_int)) %>% 
  hc_add_series(data = reg_fit, type = "line", hcaes(x = debt_ratio, y = .fitted)) %>%
  hc_xAxis(title = list(text = "Debt Ratio")) %>%
  hc_yAxis(title = list(text = "R&D to Sales Ratio"))

Q6

Creating a face plot is also not straightforward. Note that a facetplot makes sense for a static visualization. In internactive visualization, you could place multiple charts side by side on a webpage easily. You may also provide users a dropdown to select the charts they want to see.

Anyway, I will first show you an alterative to the faceplot in Q6 where we are comparing only 2 companies.

Note that I have changed the tooltip quite a bit. Try to understand this code. The last function in the code adds the labels to the bars.

d1 %>% 
  filter(tic %in% c("META", "NVDA")) %>% 
  filter(fyear >= 2010) %>% 
  hchart("column", hcaes(x = as.character(fyear), y = oibdp, group = conm, name = as.character(fyear))) %>%
  hc_colors(c("#5cc9f5", "#b131a2")) %>%
  hc_xAxis(title = list(text = "Fiscal Year")) %>%
  hc_yAxis(title = list(text = "Profits in $ million")) %>%
  hc_legend(align = "center", verticalAlign = "top", layout = "horizontal", symbolRadius = 0) %>%
  hc_tooltip(headerFormat = "",
             pointFormat = 'Year: {point.name}<br><span style="color:{point.color}">\u2714</span> {series.name}: <b>{point.y}</b><br/>') %>%
  hc_plotOptions(series = list(dataLabels = list(enabled = TRUE, format = "{y}")))

NA

Back to creating a facet plot. The solution is to create two separate plots and then arrange them using an obscure fuction called hw_grid() in highcharter. This function is not ported from the JS library. Instead the developer wrote this for R users.

Anyway, the method is to create separate plots for Meta and nvidia and then put them in one column. To save typing, I used map function from purrr package to apply tickers individually to the same plotting function. It creates 2 plots and saves them as a list named charts. Next, I use hw_grid to organize these two vertically. In other words, we can’t create facetplots as easily as we did in ggplot2.

tickers = c("META", "NVDA")
colors = list(META = "#5cc9f5", NVDA = "#b131a2")

charts <- map(tickers, function(x) {
  
  .dt = d1 %>% filter(tic == x)
  
  hchart(.dt, 
         type = "column", 
         hcaes(x = as.character(fyear), y = oibdp, name = fyear))   %>%
    hc_colors(colors[[x]]) %>%
    hc_xAxis(title = list(text = "Fiscal Year")) %>%
    hc_yAxis(title = list(text = "Profits in $ million")) %>%
    hc_title(text = .dt$conm[1]) %>%
    hc_legend(enabled = FALSE) %>%
    hc_tooltip(headerFormat = "",
               pointFormat = 'Year: {point.name}<br><span style="color:{point.color}">\u2714</span> {series.name}: <b>{point.y}</b><br/>') %>%
    hc_plotOptions(series = list(dataLabels = list(enabled = TRUE, format = "{y}")))
})

hw_grid(charts, ncol = 1, rowheight = 600)