Customizing a diverging palette heatmap

The default color scheme used by Seaborn's heatmap() doesn't give the value of 0 any special treatment. This is fine for instances when 0 isn't special for the variable you're visualizing but means you will need to customize the palette when 0 is special, such as when it represents a neutral value.

For this visualization, you want to compare all the cities against the average pollution value for CO in November 2015. (As is provided in the DataFrame nov_2015_CO).

To do this, use a heat map to encode the number of standard deviations away from the average each city's CO pollution was for the day. You'll need to replace the default palette by creating your own custom diverging palette and passing it to the heatmap and informing the function what your neutral value is.

This exercise is part of the course

Improving Your Data Visualizations in Python

Exercise instructions

Pass the diverging palette to your plot.
- Note that the sns.heatmap() function in Seaborn does not have a palette argument but instead requires the cmap argument.
Add your neutral value to the heat map by specifying the center argument.
Set the upper and lower boundaries to the color bar to -4 and 4 to make the legend symmetric.

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

# Define a custom palette
color_palette = sns.diverging_palette(250, 0, as_cmap = True)

# Pass palette to plot and set axis ranges
sns.heatmap(nov_2015_CO,
            ____ = ____,
            ____ = ____,
            vmin = ____,
            vmax = ____)
plt.yticks(rotation = 0)
plt.show()

Edit and Run Code

This exercise is part of the course

Improving Your Data Visualizations in Python

IntermediateSkill Level

4.7+

Start Course for Free

How do you show all of your data while making sure that viewers don't miss an important point or points? Here we discuss how to guide your viewer through the data with color-based highlights and text. We also introduce a dataset on common pollutant values across the United States.

Exercise 1: Highlighting data Exercise 2: Hardcoding a highlight Exercise 3: Programmatically creating a highlight Exercise 4: Comparing groups Exercise 5: Comparing with two KDEs Exercise 6: Improving your KDEs Exercise 7: Beeswarms Exercise 8: Annotations Exercise 9: A basic text annotation Exercise 10: Arrow annotations Exercise 11: Combining annotations and color

Color is a powerful tool for encoded values in data visualization. However, with this power comes danger. In this chapter, we talk about how to choose an appropriate color palette for your visualization based upon the type of data it is showing.

Exercise 1: Color in visualizations Exercise 2: Getting rid of unnecessary color Exercise 3: Fixing Seaborn's bar charts Exercise 4: Continuous color palettes Exercise 5: Making a custom continuous palette Exercise 6: Customizing a diverging palette heatmap

Current Exercise

Exercise 7: Adjusting your palette according to context Exercise 8: Categorical palettes Exercise 9: Using a custom categorical palette Exercise 10: Dealing with too many categories Exercise 11: Coloring ordinal categories Exercise 12: Choosing the right variable to encode with color

Uncertainty occurs everywhere in data science, but it's frequently left out of visualizations where it should be included. Here, we review what a confidence interval is and how to visualize them for both single estimates and continuous functions. Additionally, we discuss the bootstrap resampling technique for assessing uncertainty and how to visualize it properly.

Exercise 1: Point estimate intervals Exercise 2: Basic confidence intervals Exercise 3: Annotating confidence intervals Exercise 4: Confidence bands Exercise 5: Making a confidence band Exercise 6: Separating a lot of bands Exercise 7: Cleaning up bands for overlaps Exercise 8: Beyond 95%Exercise 9: 90, 95, and 99% intervals Exercise 10: 90 and 95% bands Exercise 11: Using band thickness instead of coloring Exercise 12: Visualizing the bootstrap Exercise 13: The bootstrap histogram Exercise 14: Bootstrapped regressions Exercise 15: Lots of bootstraps with beeswarms

Often visualization is taught in isolation, with best practices only discussed in a general way. In reality, you will need to bend the rules for different scenarios. From messy exploratory visualizations to polishing the font sizes of your final product; in this chapter, we dive into how to optimize your visualizations at each step of a data science workflow.

Exercise 1: First explorations Exercise 2: Looking at the farmers market data Exercise 3: Scatter matrix of numeric columns Exercise 4: Digging in with basic transforms Exercise 5: Exploring the patterns Exercise 6: Is latitude related to months open?Exercise 7: What state is the most market-friendly?Exercise 8: Popularity of goods sold by state Exercise 9: Making your visualizations efficient Exercise 10: Stacking to find trends Exercise 11: Using a plot as a legend Exercise 12: Tweaking your plots Exercise 13: Cleaning up the background Exercise 14: Remixing a plot Exercise 15: Enhancing legibility Exercise 16: Congrats!