Visualizations are an important part of any data storytelling. These can be simple graphs or interactive data visualizations, multiple dimension views, maps, or animation.

Choosing an appropriate visualization depends on your audience’s knowledge level, data size, format, dynamics, and purpose. The biggest challenge with picking an appropriate data chart is to present the data value in the most easily recognizable way. 

Types of visualization

To start with visualizations, you have to understand what type of problem you are describing - what are you trying to find in your data? 

1. Comparison and Ranking

2. Relationship and Correlation

3. Distribution and Evolution

4. Composition and Flow

Each of those categories has its own set of graphs depending on what format and type your data is - quantitative or qualitative.

Comparison and Ranking

Comparison or Ranking is the most common type of visualization for data storytelling. You can compare events over time or with each other. Common methods include table charts, columns, and bar charts. Over time comparisons include line charts. You would use ranking or comparison charts to show the lowest and highest values in the data. 

Relationship and Correlation

Relationship visualizations can have two or more variables, and the best representation would be done via scatter or bubble chart. You would use these to show data clusters, outliers, or correlations. 

Distribution and Evolution

Distribution and Evolution charts work best for representing quantitative values. This chart choice also depends on how many variables you have. These can be histograms (column and line), scatters, box plots, or 3D area chart (for 3 variables).

Composition and Flow

Composition and Flow visuals can be static or dynamic. Static charts are waterfalls graphs, pie charts, or stacked column charts. Dynamic depend on how many relative differences you aim to display for how many periods of time. You would use this type of visualization to see the relative value or absolute difference - either percentage of total or value of total.  

Images are taken from here and here.

If you feel lost, follow this guide to pick the right visual for your story:

Taken from here.

Each of those problem types described above could have different visualization techniques: charts, plots (scatter, bubble), maps (heat maps, cartograms), diagrams (multidimensional, trees), and matrixes.

🔮 My recommendations for the best practices for data visualization: 

• For displaying negative numbers or comparing 10 or more items, use a bar graph instead of a column chart.

• Don’t use pie charts for more than 4 categories. Use a table instead. 

• For continuous data, avoid plotting more than 3-4 lines in a line graph.

• If you are working with multiple datasets and have different variables, instead of multiple charts, you can make one dual-axis chart with 1 X-axis and 2 Y-axes. For example, you could combine a line chart and a column chart in one graph to show the relationship between different variables. 

• Use colors to convey a story (green/red - good progress or slow). Avoid bright saturated color choices. Read an amazing guide on how to pick more beautiful colors for your data visualization.

🔥 Best examples of data visualization

• Job Market Tracker from WSJ. Very easy to read and intuitive. Smart use of colors and scale.

• Coronavirus tracker from San Francisco Chronicle. Simple and clear graphs without any additional noise or labels. Smart use of dual-axis charts that I mentioned above. 

• Another amazing set of interactive visuals -  Simulating an epidemic.

• Get inspired by the gallery of most latest sophisticated 3D charts.

• YouTube Trending Videos Analysis -  a thorough deep-dive in Python with great visualizations.   

📚 Helpful materials on making visualizations 

There are many tools for data visualization, from Tableau or PowerBI to Plotly, from Sisense or Grafana to R, Python, or JavaScript libraries. 

Here is a round-up of my favorite How To resources on how to build a chart or design a graph:

Python:

• Getting started with visualizations in Pandas

• Data Visualization Using Python’s Matplotlib and Seaborn Packages

• Data Visualization Tutorial Using Seaborn

R:

• Step by step guide on how to plot Time Series in R

• Improving GGPlot visualizations  

• R Graph Gallery

Other:

• Getting started with Plotly and Dash

• Building charts in JS

• How to design a dashboard - a complete dashboard development guide with a checklist.

  ---

Make sure to subscribe to learn more about data analysis and never miss an update: