Click the Original, Code and Reconstruction tabs to read about the issues and how they were fixed.

Original


Source: Timeline of Retail and E-Commerce Sales within the U.S. Retail Industry, U.S. Census Bureau (2020).


Objective

The objective of the original data visualisation is to showcase the contribution of e-commerce sales as a percentage of the total retail sales made in each given year over a specified duration of time. The aim is to gain insights on whether there is a boom in e-commerce retail sales in the past decades. This information is beneficial to companies who operate within the U.S. retail industry. Information interpreted from the visualisations will benefit any existing player within the U.S. retail industry or any new potential entrants looking to enter the U.S. market with their short-term and long-term business strategies. To name a few of the target audiences within the above mentioned market spaces includes but are not limited to company strategists such as analysts, C-suite executives and stakeholders.

The visualisation chosen had the following three main issues:

  • The alignment of the exploding sequenced pie chart does not clearly convey the percentage change in growth of the contribution of the boom of e-commerce sales out of the total retail sales over a given time period of two decades. For instance, the different sizes of the colour red’s hue and saturation portions on the chart is meant to showcase the percentage of e-commerce sales out of the total retail sales made for each given year; however it is evidently confusing to the reader for two main reasons which are:

  1. the visual bombardment due to saturation of the colours blue and red

  2. sizes of the red and blue portions of the chart are not measured to scale. Since the chart is not measured a scale, it makes it difficult for the reader to understand, interpret and visualise the percentage growth over the given period of time. As a result of this, the data visualisation seems to be misleading which means it is deceptive in nature and less effective in conveying the objective of the visualisation. Therefore, a bar graph would be more indicative when comparing percentages of e-commerce sales in different years in either an ascending or descending order. Moreover, adding grids to the bar graph will be helpful as it will help with highlighting the small percentage changes, assuring accuracy in reading the graph.

  • Although the visualisation has percentage of e-commerce sales out of the total retail sales mentioned, the absence of an axis makes it more difficult for the human eye to elucidate the objective of the chart. This is because the human eye is better trained and equipped to interpret visualisations along graphs plotted against the y- and x- axes. Thus, the absence of axes creates a bias and leads to misinformation that is misleading judgement and/or interpretations.

  • The objective of the visualisation revolves around the percentage of e-commerce boom of the total retail sales made each year, which is represented by the various shades of red as seen in the visualisation. There are two main issues with this colour representation which are as follows:

  1. the four shades of red ranging from light red which comes off as pink to the naked eye to a stark, bright dark red colour. This is an issue as it can impair the ability of any human being that suffers from red colour blindness. This increases the probability of misreading the data visualisation resulting in making wrong conclusions; which is vital as it is makes up for the essence of the aim of the visualisation. This is why it is essential to keep in mind a colourblind-safe colour palette when producing visualisations. Moreover, red-blind readers perceive the colour as blue and/or olive whilst blue-blind readers perceive the colour as teal and/or pink. This, with regards to our visualisation will create immense confusion given the interchange of colours due to colour blindness, therefore resulting in misleading information thus defeating the purpose of the objective.

  2. Since the primary focus is the data presented in shades of the colour red in the visualisation, the different shades of blue showcasing total retail sales is not required. It results in unnecessary information and colour explosion, especially where it is not essential or required with regards to the objective of the visualisation.

References

Code

The following code was used to fix the issues identified in the original.

library(dplyr)
library(ggplot2)

retailsales <- data.frame(Year = c("2020", "2019", "2018", "2017", "2016", 
                                   "2015", "2014", "2013", "2012", "2011"),
                          Percentage_of_Ecommerce_Total_Retail_Sales = c(14.0, 10.9, 9.8, 9.0, 
                                                                         8.2, 7.3, 6.5, 5.9, 5.4, 4.9))
View(retailsales)

retailsales1 <- retailsales[order(retailsales$Year),]
View(retailsales1)

plot1 <- ggplot(data = retailsales1, aes(group = 1, x = Year,y = Percentage_of_Ecommerce_Total_Retail_Sales))
plot1 <- plot1 + geom_bar(stat= "identity", fill= "purple", colour= "black") +
      geom_text(aes(label = paste(Percentage_of_Ecommerce_Total_Retail_Sales, "%", sep='')), 
                nudge_y = 1, nudge_x = 0.05) + 
      labs(
         title = "Change in Growth of E-commerce Retail Sales in the U.S. Over a Decade", 
             y = "% of Ecommerce Retail Sales") + theme(plot.title = element_text(hjust = 0.5),   
             panel.background = element_rect(fill = "white", colour = "black", linetype="solid", size=1),
             panel.grid.major = element_line(size = 0.5, linetype = 'solid', colour = "grey"),
             panel.grid.minor = element_line(size = 0.25, linetype ='solid', colour ="grey")) + 
             scale_y_continuous(limits = c(0,16))

Data Reference

  • SSSD, Advance Monthly Retail, Monthly Retail, and Quarterly E-Commerce 2019, Monthly Retail Trade, Main Page - US Census Bureau, Census.gov, viewed 19 September 2022, https://www.census.gov/retail/index.html

Reconstruction

The following plot fixes the main issues in the original.