KDE Plot

Kernel density estimation (KDE) plot is a plot used to visualize the probability density function estimation. It is in a way similar to a histogram which we discussed in the previous section, however, the KDE plot is a continuous curve, not a set of bars, and is based on all of the data points rather than the intervals. Let’s have a look at an example of a KDE plot:

As you can see, here we have a histogram combined with a KDE plot (orange curve). This combination gives us a much clearer probability density function approximation than a single histogram.

With seaborn creating a KDE plot is as simple as it gets, since there is a special kdeplot() function. Its most important parameters data, x and y work the same way as in the countplot() function.

First Option

We can simply set only one of these parameters via passing a sequence of values. Here is an example to clarify everything:

import pandas as pd

import matplotlib.pyplot as plt

import seaborn as sns

url = 'https://content-media-cdn.codefinity.com/courses/47339f29-4722-4e72-a0d4-6112c70ff738/weather_data.csv'

# Loading the dataset with the average yearly temperatures in Boston and Seattle

weather_df = pd.read_csv(url, index_col=0)

# Creating a KDE plot setting only the data parameter

sns.kdeplot(data=weather_df['Seattle'], fill=True)

plt.show()


              123456789
            
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
url = 'https://content-media-cdn.codefinity.com/courses/47339f29-4722-4e72-a0d4-6112c70ff738/weather_data.csv'
# Loading the dataset with the average yearly temperatures in Boston and Seattle
weather_df = pd.read_csv(url, index_col=0)
# Creating a KDE plot setting only the data parameter
sns.kdeplot(data=weather_df['Seattle'], fill=True)
plt.show()

We only set the value for the data parameter passing a Series object and use the fill parameter to fill in the area under the curve (it is not filled in by default).

Second Option

It is also possible to set a 2D object like a DataFrame for data and a column name (or a key if the data is a dictionary) for x (vertical orientation) or y (horizontal orientation):

import pandas as pd

import matplotlib.pyplot as plt

import seaborn as sns

url = 'https://content-media-cdn.codefinity.com/courses/47339f29-4722-4e72-a0d4-6112c70ff738/weather_data.csv'

weather_df = pd.read_csv(url, index_col=0)

# Creating a KDE plot setting both the data and x parameters

sns.kdeplot(data=weather_df, x='Seattle', fill=True)

plt.show()


              12345678
            
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
url = 'https://content-media-cdn.codefinity.com/courses/47339f29-4722-4e72-a0d4-6112c70ff738/weather_data.csv'
weather_df = pd.read_csv(url, index_col=0)
# Creating a KDE plot setting both the data and x parameters
sns.kdeplot(data=weather_df, x='Seattle', fill=True)
plt.show()

We achieved the same results passing the whole DataFrame as the data parameter and the column name for the x parameter.

By the way, the KDE plot we created has a characteristic bell curve and closely resembles the normal distribution with the mean of approximately 52°F.

In case you want to explore more about the kdeplot() function, feel free to refer to its documentation.

Oppgave

Swipe to start coding

Use the correct function to create a KDE plot.
Use countries_df as the data for the plot (the first argument).
Set 'GDP per capita' as the column to use and the orientation to horizontal via the second argument.
Fill in the area under the curve via the third (rightmost) argument.