Questions

• Is the message clear?
• Are the colors helpful?
• Do the data support the intended story?
• Is the 3D effect of the line justified?
• Is the Netflix axis equivalent to the Tesla axis?
• Assuming the time frame is the same for both stocks, what is the approximate slope for NFLX and for TSLA? Is that clear from the visualization?

The Good, the Bad, and the Ugly

We can summarize some thoughts about the original graphic before starting on an improved version.

good<-c("complementary colors", "strong visual impact")
bad<-c("scale", "distracting graphics")
ugly<-c("3D lines", "misleading")

df<-data.frame(good,bad,ugly)
names(df)<-c("The Good", "The Bad", "The Ugly")

kable(df)

The Good	The Bad	The Ugly
complementary colors	scale	3D lines
strong visual impact	distracting graphics	misleading

Analysis

Our goal is to recreate the data visualization but with data integrity as the priority. We also will attempt to recreate the overlapped data as presented earlier to try and determine the accuracy of the plot.

First, we use library(quantmod)66 “The quantmod package for R is designed to assist the quantitative trader in the development, testing, and deployment of statistically based trading models.” (quantmod documentation) to import stock data.

library(quantmod)

#set dates for stock price retrieval
start <- as.Date("2011-06-01")
end <- as.Date("2013-11-01")

start2 <- as.Date("2019-01-01")
end2 <- as.Date("2019-07-01")

#import data using the dates above
getSymbols("NFLX", src = "yahoo", from = start, to = end)

## [1] "NFLX"

getSymbols("TSLA", src = "yahoo", from = start2, to = end2)

## [1] "TSLA"

#plot(NFLX[,"NFLX.Close"], main = "NFLX Close")
#plot(TSLA[,"TSLA.Close"], main = "TSLA Close")

#creates an xts time series
nf <- as.xts(data.frame(NFLX = NFLX[,4]))
ts <- as.xts(data.frame(TSLA = TSLA[,4]))

#fortify will convert xts to a dataframe with Index as a column
nff <- fortify(nf)
tsf <- fortify(ts)

#convert Index columns in each dataframe to date type 
#(scale_x_date below only works with date types)
nff$Index<-as.Date(nff$Index)
tsf$Index<-as.Date(tsf$Index)

Use ggplot() to plot a more honest comparison with accurate scales for time and price. Note the use of ‘scale_x_date’ for the X axis77 Be careful when using dates! Check the data type using class(myVariable). If the data are not ‘Date’ type, convert using as.Date(myVariable).. The axis text angle is modified after the theme_minimal(). Highlighting88 Highlighting can help draw user attention to desired features. Color matching can help reinforce associations. is implemented here to draw attention to the regions of interest in each stock data set. Notice that the compared regions represent very little of the Netflix data, but a large proportion of the Tesla data.

ggplot() +
  geom_rect(data=NULL,aes(xmin=as.Date("2011-06-01"),xmax=as.Date("2011-10-20"),ymin=-Inf,ymax=Inf), fill="#00000011")+
  geom_rect(data=NULL,aes(xmin=as.Date("2019-01-02"),xmax=as.Date("2019-05-24"),ymin=-Inf,ymax=Inf), fill="#FF000011")+
  geom_line(data=nff, aes(x=Index, y=NFLX.Close, color="Netflix"))+
  geom_line(data=tsf, aes(x=Index, y=TSLA.Close, color="Tesla"))+
  scale_x_date(date_labels = "%b %y",date_breaks = "6 months")+
  scale_y_continuous(breaks = seq(0, 400, by=50), name="Closing Price ($USD)")+
  theme_minimal()+
  theme(axis.text.x = element_text(angle = 45, hjust = 1))+
  scale_colour_manual(
  values = c("#333333", "#FF0000"),
  labels = c("Netflix", "Tesla"))+
  labs(color="Company")

Now we can force an overlap by scaling TSLA down. We can also add an additional Y axis for the Tesla data. Note the use of ‘sec.axis’ to create a scale equal to the existing Y values times 10. This converts the scaled-down data values back to the correct value for the axis. 99 Is this figure easier or more difficult to interpret?

#add ID columns using tibble function rowid_to_column
nff <- rowid_to_column(nff, "ID")
tsf <- rowid_to_column(tsf, "ID")

#merge the dataframes by ID number
mergeDF<-merge(nff,tsf,by="ID")

newPlot<-ggplot() +
  geom_line(data=nff, aes(x=ID, y=NFLX.Close, color="Netflix"))+
  geom_line(data=tsf, aes(x=ID, y=TSLA.Close/10, color="Tesla"))+
  scale_y_continuous("Closing Price ($USD, NFLX)",
                     breaks = seq(15, 50, by=5), 
                     sec.axis = sec_axis(~.*10, name = "Closing Price ($USD, TSLA)",
                                         breaks = seq(150, 500, by=50)))+
  #scale_y_continuous(name="Closing Price ($USD, NFLX)", breaks = seq(15, 40, by=5))+
  #scale_y_continuous(sec.axis = sec_axis(~.*10, name = "Closing Price ($USD, TSLA)",breaks = seq(150, 400, by=50)))+
  scale_colour_manual(values = c("#CCCCCC","#FF3300"),name = "Company")+
  scale_x_continuous(name = "Day")+
  theme_minimal()+
  theme(axis.title.y      = element_text(margin = margin(t = 0, r = 10, b = 0, l = 0)))+
  theme(axis.title.y.right= element_text(margin = margin(t = 0, r = 0, b = 0, l = 10)))+
  theme(axis.text.y.right = element_text(color = "red"))+
  labs(title = "TECH STOCK THROWBACK?",
              subtitle = "Comparison of closing price (NFLX: 2011-06-01 to 2011-10-20; TSLA: 2019-01-02 to 2019-05-24)",
              caption = "Data Source: Yahoo")
newPlot

We can restrict values to 100 days for a more direct comparison.

⊕Zooming in on the important message can help with clarity. The original plot suggests that perhaps Tesla will rebound like Netflix. Is there any reason to believe this given the current trend? Why or why not?

#select first 100 rows for comparison
selectedDF<-slice(mergeDF, 1:100)

Here’s a focused version of the data. Does the Tesla trend remind you of Netflix as suggested by Eddie Yoon? Why or why not?

ggplot() +
  geom_line(data=selectedDF, aes(x=ID, y=NFLX.Close, color="Netflix"))+
  geom_line(data=selectedDF, aes(x=ID, y=TSLA.Close/10, color="Tesla"))+
  scale_y_continuous("Closing Price ($USD, NFLX)",
                     breaks = seq(15, 40, by=5), 
                     sec.axis = sec_axis(~.*10, name = "Closing Price ($USD, TSLA)",
                                         breaks = seq(150, 400, by=50)))+
  #scale_y_continuous(name="Closing Price ($USD, NFLX)", breaks = seq(15, 40, by=5))+
  #scale_y_continuous(sec.axis = sec_axis(~.*10, name = "Closing Price ($USD, TSLA)",breaks = seq(150, 400, by=50)))+
  scale_colour_manual(values = c("#CCCCCC","#FF3300"),name = "Company")+
  scale_x_continuous(name = "Day")+
  theme_minimal()+
  theme(axis.title.y      = element_text(margin = margin(t = 0, r = 10, b = 0, l = 0)))+
  theme(axis.title.y.right= element_text(margin = margin(t = 0, r = 0, b = 0, l = 10)))+
  theme(axis.text.y.right = element_text(color = "red"))+
  labs(title = "TECH STOCK THROWBACK?",
              subtitle = "Comparison of closing price (NFLX: 2011-06-01 to 2011-10-20; TSLA: 2019-01-02 to 2019-05-24)",
              caption = "Data Source: Yahoo")