Source: https://drive.google.com/file/d/1ZNHoCyafmcNOL_E4kb86yiYIXnol2FPs/view?usp=sharing
Data Visualization
Histograms
Information about a single set of numbers
Range of possible height values is easily visualized
Densityplot
Information about a single set of numbers
Demonstrates the distribution of the data (a smoothed version of a histogram)
Scatterplot
Relationship between two numerical variables
Barplot
Count of values within a categorical variable
Boxplot
Summary of numerical values across categories
Line plot
Quantitative trend over time
Grouped bar chart
Count information about two categorical variables
Stacked bar chart
Counts/proportions across categorical variables
Tables
Effective ways to display data summaries
Visualization best practices
- Choose the right type of visualization
- Label your axes
- Make sure the text size is big enough
- Make sure your numbers add up
- Make sure the numbers and graphics represent the data
- Make comparisons easy on your readers
- Use y axes that start at 0 for barplots
- Keep it simple
Color Considerations
Use color to highlight your point. And use it well.
- Be mindful when choosing colors.
- Gradients are for continuous values. Distinct colors are for categories.
- Color consistency helps viewers.
- Gray is your friend.
- Choose intuitive colors.
Other Considerations
- Allow viewer to make comparison top to bottom.
- Order rows logically.
- Order columns logically.
- Limit the number of rows and columns.
- Include informative labels.
- Be mindful of significant digits.
- Include a good caption.
- Include the source of data.
- Format table so it can be quickly understood.
Summary: Visualization Best Practices
- Choose the right type of visualization
- Use appropriate colors
- Label your axes
- Make sure everything is big enough.
- Make sure numbers add up
- Make sure plot reflects the data
- Make comparisons easy on the reader
- Don’t deceive viewers
- Keep it simple
Why effective data communication matters
- It’s often the only thing your coworkers/bosses see.
- It can set your work apart from others’.
- It helps show off the awesome stuff you’ve done.
- Cognitive load is a thing.
Exploratory vs. Explanatory
Creating good visualizaitons is an iterative process
Exploratory visualization
- to take a quick look at your data
- generated during EDA
Explanatory visualization
- used in a presentation, report, or email to your company
- time-consuming to generate; reserved for plots that tell a story
Creating Visuals for Communication (aka Explanatory Data Visualizations)
- What’s your point?
- How can you emphasize your point in a chart?
- What does the final chart show exactly?
What is your point?
Start by plotting some data (an exploratory plot)
Use the title to tell the viewer what conclusions are to be drawn
Use color effectively
How can you emphasize your point in a chart?
Gray is a data visualizer’s best friend
Use boxes and text to highlight important points in your visualization
What does the final chart show exactly?
Make sure viewers know what the numbers represent and units
Directly label the data (rather than rely on a legend)
Always include your source
Aim to improve your data:ink ratio
Everything on your page should serve a purpose. If it doesn’t, remove it! (Declutter)