course_model

Visualize Data with Seaborn

This tutorial shows common charts made with the Seaborn and Matplotlib libraries.

Outcomes:

Create a bar plot
Create a count plot
Create a histogram
Create a box plot
Create a scatter plot
Create a pair plot
Customize chart titles and axis labels
Change chart styles and color palettes
Use pre-built code to quickly visualize a dataset

Links:

template

Optional Reading

Handbook: Seaborn

Bar Plot

Use bar plots to compare means or medians across categories.

Required:

data = dataframe: your dataset
x = 'fieldname of category': categorical variable
y = 'fieldname of quantitative': quantitative variable (or switch to x for a horizontal bar plot)

Optional:

hue = 'fieldname of category': adds color grouping for each bar
estimator = np.mean: function to compute the value to be plotted (default is mean, but you can use np.median, np.sum, etc.)
errorbar = ('ci', 95): confidence interval for the estimate (default is 95%)

# Barplot Example
import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np

df_penguins = sns.load_dataset("penguins")

# Bar plot: average body mass by species
sns.barplot(data=df_penguins, y="species", x="body_mass_g", hue="sex", errorbar = ('ci', 50), estimator=np.median)
plt.title("Bar Plot: Average Body Mass by Species")
plt.show()

png

Count Plot

Similar to bar plot, but for counting occurrences of categories.

# Count plot: frequency of species

sns.countplot(data=df_penguins, x="species")
plt.title("Count Plot: Number of Penguins by Species")
plt.show()

png

Line Plot

Show trends over time

Required:

data = dataframe: your dataset
x = 'fieldname of quantitative': quantitative variable for x-axis
y = 'fieldname of quantitative': quantitative variable for y-axis

Optional:

hue = 'fieldname of category': adds color grouping for each line. Disables error bars by default.
estimator = np.mean: function to compute the value to be plotted (default is mean)
errorbar = ('ci', 95): None or a tuple with the confidence interval for the estimate (default is 95%)

# Lineplot
import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np

# Load example dataset
df_flights = sns.load_dataset("flights")

# Line plot showing passengers per year
sns.lineplot(data=df_flights, 
                x="year", y="passengers", 
                errorbar=("ci", 80), 
                estimator=np.median) # or np.mean
plt.title("Line Plot: Passengers Over Time")
plt.show()

png

Scatter Plot

Show relationship between two quantitative variables.

Required:

data = dataframe: your dataset
x = 'fieldname of quantitative': quantitative variable for x-axis
y = 'fieldname of quantitative': quantitative variable for y-axis

Optional:

hue = 'fieldname of category': adds color grouping for each point.
size = 'fieldname of size': adds size grouping for each point.

# Scatterplot
import seaborn as sns
import matplotlib.pyplot as plt

# Sample data
df_penguins = sns.load_dataset("penguins")

# Add third variable (species as hue, body_mass as size)
sns.scatterplot(data=df_penguins,
                x="bill_length_mm", y="bill_depth_mm",
                hue="species", 
                size="body_mass_g",
                alpha=0.25)  # very helpful when points overlap, goes from 0 to 1 (transparent to opaque)
plt.title("Scatter Plot with Hue & Size")
plt.show()

png

Box Plots

Show medians, quartiles, and outliers.

Required:

data = dataframe: your dataset
x = 'fieldname of category': categorical variable
y = 'fieldname of quantitative': quantitative variable

Optional:

hue = 'fieldname of category': adds color grouping for each box

# Boxplot
import seaborn as sns
import matplotlib.pyplot as plt

df_penguins = sns.load_dataset("penguins")

# Box plot: flipper length across species
sns.boxplot(data=df_penguins, x="species", y="flipper_length_mm", hue="sex")
plt.title("Box Plot: Flipper Length by Species")
plt.show()

png

Violin Plots

Similar to boxplot, but shows more detail about distribution shape.

# Violin plot
import seaborn as sns
import matplotlib.pyplot as plt

df_penguins = sns.load_dataset("penguins")

# Box plot: flipper length across species
sns.violinplot(data=df_penguins, x="species", y="flipper_length_mm", hue="sex")
plt.title("Violin Plot: Flipper Length by Species")
plt.show()

png

Pair Plots

Show relationships between all pairs of quantitative variables.

Required:

data = dataframe: your dataset

Optional:

hue = 'fieldname of category': adds color grouping for each point
kind = 'scatter': type of plot to show in the off-diagonal (default is scatter, but can also be ‘reg’ for regression line, ‘kde’ for kernel density estimate, etc.)
alpha: transparency level for points in scatter plots (default is 1, fully opaque). You can only use this with the following line (and you must use the reg kind argument):
```
sns.pairplot(data=df, hue='category', kind='reg', plot_kws={'alpha': 0.25})
```

# Pairplot
import seaborn as sns
import matplotlib.pyplot as plt

df_penguins = sns.load_dataset("penguins")

# Box plot: flipper length across species
sns.pairplot(data=df_penguins, 
        hue="species",
        kind="reg",
        height=1.5, 
        plot_kws={'scatter_kws': {'alpha': 0.1}})
plt.title("Pair Plot: Penguins Dataset")
plt.show()

png

Sample Code

The below code is useful for a quick overview of all variables in a dataset. It creates a plot for each variable. Numbers are shown with a histogram and boxplot, and text with a count plot.

Update it by updating df to your dataframe of choice.

import matplotlib.pyplot as plt
import seaborn as sns

df = sns.load_dataset("penguins")

number_columns = df.select_dtypes(include=['number']).columns.tolist()
text_columns = df.select_dtypes(include=['string', 'category']).columns.tolist()

# Print a histogram and boxplot for each numeric column
for col in number_columns:
    plt.figure(figsize=(10, 4))
    plt.subplot(1, 2, 1)
    sns.histplot(df[col])
    plt.title(f'Histogram of {col}')
    
    plt.subplot(1, 2, 2)
    sns.boxplot(x=df[col].dropna())
    plt.title(f'Boxplot of {col}')
    
    plt.tight_layout()
    plt.show()

# Print a chart showing the distribution of values for each text column
for col in text_columns:
    plt.figure(figsize=(6, 4))
    sns.countplot(y=df[col], order=df[col].value_counts().index)
    plt.title(f'Value counts of {col}')
    plt.show()

png