Summary

The webpage provides an in-depth guide on various data visualization techniques, particularly focusing on the visualization of the COVID-19 dataset to demonstrate their effectiveness in data analysis.

Abstract

The article "How to ace Data Visualization" emphasizes the importance of visual content in data science, illustrating how the human brain processes visual information more efficiently than text or numbers. It showcases a range of visualization techniques, including bar charts, pie charts, treemaps, word clouds, and geographical maps, using the COVID-19 dataset as a practical example. The author highlights the need to choose appropriate visualization methods based on the type of data (categorical or continuous) and the context of the report, such as the number of categories or the size of the dataset. The article also touches on the use of visualization tools and the combination of visuals with text and numbers for compelling storytelling in data reports. The visualizations presented help to quickly convey crucial insights, such as the distribution of cases, deaths, and recoveries across different countries, and the effectiveness of vaccination efforts over time.

Opinions

The author believes that data visualization is a powerful tool for quickly identifying patterns and understanding complex datasets.
Visual representations, such as diagrams and infographics, are considered more effective for retaining information and engaging audiences than textual or numerical data alone.
The popularity of data visualization is reflected in the abundance of visualization tools currently available in the market.
The author suggests that the choice of visualization technique depends on individual preferences, the nature of the data, and the report's requirements.
Pie charts are critiqued for potentially being misleading when comparing different categories, with the author recommending stacked normalized bar charts as a better alternative in some cases.
Geographical maps and treemaps are praised for their ability to create engaging and memorable content, making them suitable for presenting data with a geographical or hierarchical aspect.
The author advocates for the use of scatter plots, line plots, and bar plots to effectively visualize continuous data and time series data, such as the progression of COVID-19 cases and vaccinations.
The article concludes by encouraging readers to explore further resources and reach out for discussions, indicating the author's commitment to community engagement and continuous learning in the field of data visualization.

Start guide

How to ace Data Visualization

Data visualization techniques are a quick way to identify patterns and understand the complex dataset. As a result, it is widely used in several industries to present the data to stakeholders. This article shows various data visualization techniques which can be helpful for your next data science project.

The human brain responds well and retains more information from simple diagrams or visual content than text or numbers. Therefore, representing a complex dataset in graphical format is an effective way to drive crucial insights and gain more information about the data. Furthermore, the popularity of data visualization techniques can be estimated through the number of visualization tools available now in the market. Many online platforms and businesses use data visualization techniques to present data as visual content (infographics), which helps them deliver crucial information quickly. The visual content registers in the brain quickly and helps create engaging content for the audience.

Figure showing a visual and textual representation of the definition of a circle.

The figure above shows a simple example of how visual representation can be processed quickly compared to text. However, this does not mean that text or numbers are less used. On the contrary, many companies use a combination of visuals, text, and numbers for their reports as a compelling storytelling method about their process, growth, profits, or losses.

Visualizing COVID-19 dataset

This article uses the COVID-19 dataset made available from the Center for System Science and Engineering (CSSE) as an example for showcasing different data visualization techniques.

Categorical values

Categorical data can be ordinal or nominal. Several data visualization techniques offer efficient and elegant ways to visualize categorical data. However, it also depends on individuals which techniques suit best for their case.

In general, the real-world dataset may contain missing values. Hence, the first step for any data-centric project involves checking the missing values. There are various ways to present missing values in a report.

Figure showing missing values in the dataset.

When the dataset has fewer columns, it is easy to use pandas; however, when the number of columns grows, a bar chart or matrix may be a better choice. The missingno library provides an easy and quick way to visualize and summarize the integrity of the dataset. When the number of categories is less, then often using numbers as indicators of frequencies are used. For example, the figure below shows the number of global COVID-19 cases.

Figure showing the number of global cases of COVID-19.

With a simple diagram, one can directly visualize the current scenario of COVID-19 cases. However, when the number of categories is large, another technique should be used instead of the above method.

When visualizing the number of cases based on countries in such cases Pie chart can be used, for example, the figure below shows the number of confirmed cases, death cases, and people recovered for the top 15 countries most affected by COVID-19.

Figure showing top 15 countries affected by COVID-19 cases.

In the figure above, each slice of pie represents the fraction of cases of a country in a group. However, comparing two different pie can be misleading. For example, comparing a pie chart of death cases to a pie chart of recoveries for the United States may conclude that deaths are more than recoveries which is not the case. Hence, for such comparisons, it is advisable to create a separate pie chart for each country.

Figure showing COVID-19 cases by countries.

The figure above shows a clear picture of all cases when created pie charts for each country. However, creating a pie chart for each country would result in too many graphs. Moreover, it is hard to display everything on a single report; hence the better visualization technique could be a stacked normalized bar chart.

Figure showing COVID-19 cases by countries — an alternative to a pie chart.

In this case, a stacked normalized bar chart seems to be a good alternative. However, a pie chart also has the advantage as it can represent data very compactly. In the case of the bar chart, the size depends on the dataset. The figure below shows the comparison between the pie chart and bar chart for 14 days incidence rate of COVID-19 cases.

Figure showing a comparison between bar chart and pie chart

In short, it depends on individual preferences and the size of the report to choose the right graph.

The treemap is also one of the commonly used visualization techniques to represent the categorical features based on the frequency. The figure below shows the confirmed cases of COVID-19.

Figure showing confirmed COVID-19 cases.

A bigger size of the box means a higher number of cases.

Wordcloud is also a great contender and an artistic way to present the dataset. The figure below shows different ways to represent the dataset.

Figure showing COVID-19 cases and population of countries

A bigger text size means a higher number. With a simple wordcloud, one can easily extract meaningful information from the dataset.

The other exciting way to visualize a dataset based on geographical location is by using the map. The figure below shows the confirmed cases by country.

Figure showing COVID-19 confirmed cases.

The bigger circle size here represents the higher number of COVID-19 cases. Thus, this kind of data visualization technique creates more engaging content and easy to remember.

Continuous values

There also exists a plethora of visualization techniques for continuous values. The most common method to visualize continuous values through a scatter plot.

Figure showing death vs. confirmed cases

The scatter plot above shows the number of confirmed cases vs. deaths, color-coded by continent. However, one can add more information to the scatter plot by using the size of the marker.

The figure above shows the number of confirmed cases vs. deaths on a log scale. The size of the marker represents the country with a large number of COVID-19 cases.

The other visualization technique for plotting the time series data is a line plot. The figure below shows the global rise in confirmed cases, deaths, and recoveries of COVID-19 cases.

Figure showing the global rise in cases and recoveries over time

One can also use the stacked or overlay mode of the bar chart to show the rise of COVID-19 cases for the U.S., India, and Brazil.

Figure showing the global rise in cases over time

These plots are a fantastic way to compare continuous values based on categories. For example, the figure below shows the daily vaccination for the countries whose confirmed cases of COVID-19 shown above.

Figure showing daily vaccination over time.

One can use line plot and bar plot to show more information. For example, the figure below shows the new cases in India with a 7 days rolling average over time.

Figure showing new cases in India over time.

That’s it.

One can use different combinations of visualization techniques to generate compelling and engaging visual content.

I hope you enjoyed this article. Any questions? Have I missed something? Please reach out on my LinkedIn or Twitter.

Further interesting articles:

Follow this Github link to access all the resources used for this article.

Cheers!

Rahul