Learn Estimating Means with Bar Plots

A barplot represents an estimate of the central tendency (usually the mean) for a numeric variable with the height of each rectangle.

Unlike a histogram which counts how many items are in a bin, a barplot calculates a statistic (like the average bill amount) for each category.

The Error Bars (Uncertainty)

The small black lines on top of each bar are called error bars. By default, they show the 95% Confidence Interval.

To customize them, we now use the err_kws (error keywords) parameter. This accepts a dictionary of settings that control the look of these lines.

Key Parameters

capsize: adds a small horizontal line ("cap") to the ends of the error bars;
err_kws: a dictionary to customize error bars;
{'color': 'black'}: sets the color;
{'linewidth': 2}: sets the thickness;
estimator: by default, it calculates the mean. You can change this to median, sum, or max (requires importing numpy).

Example

Here is a barplot showing the average total bill with customized red error bars.


              12345678910111213141516171819
            
import seaborn as sns
import matplotlib.pyplot as plt

# Load dataset
df = sns.load_dataset('tips')

# Create a barplot
sns.barplot(
    data=df,
    x='day',
    y='total_bill',
    hue='sex',
    capsize=0.1,
    # New way to style error bars:
    err_kws={'color': 'red', 'linewidth': 2},
    palette='pastel'
)

plt.show()

Task

Swipe to start coding

Calculate and visualize the average total bill for different days, comparing smokers and non-smokers.

Set the style to 'ticks' to remove the grid.
Create a barplot using the tips dataset (df):
- Map 'day' to x and 'total_bill' to y.
- Group by 'smoker' using hue.
- Set the error bar capsize to 0.1.
- Change the error bar color to 'pink' using the err_kws dictionary (e.g., {'color': 'pink'}).
- Set the bar outline linewidth to 2.5.
- Use the 'magma' palette.
Display the plot.

Solution

Everything was clear?

Thanks for your feedback!

Section 1. Chapter 13

single

Ask AI

Ask anything or try one of the suggested questions to begin our chat

Swipe to show menu