Distinctive Analytics

Matplotlib is at the heart of Seaborn Library’s data point visualizations. This library gives highly statistically informative graphics functionality to Seaborn.

Installation

Following are short instructions for installing seaborn.

# command prompt
pip install seaborn
# jupyter lab / collab
!pip install seaborn

You can use these commands on the command prompt. Use the same command on Jupyter or Collab in the code cell.

Importing

While importing Seaborn, you should also import the Matplotlib and pyplot objects for visualizing the plots. In the following image, we import seaborn, numpy, and pandas for visualizations, mathematical functions, and dataset handling.

import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt

KDE plots with Distribution Densities

KDE plots are used to show the data distribution. In the following code, we use artificially produced data to plot with KDE.

Python

Python

Python

data = np.random.multivariate_normal([2, 3], [[4, 6], [7, 9]],size=2000)
data = pd.DataFrame(data, columns=['A', 'B'])
sns.set(rc={'figure.figsize':(11.7,8.27)})
sns.kdeplot(data['A'], shade=True)
sns.kdeplot(data['B'], shade=True)

Distplot

Distplot can help with observing histograms with trend lines. Distplot is useful for studying continuous data.

Python

Python

Python

sns.set(rc={'figure.figsize':(11.7,8.27)})
sns.distplot(data['A']);
sns.distplot(data['B']);

KDE plot 2D

We can also plot the data with KDE with a 2-factor distribution plot.

Python

Python

Python

sns.kdeplot(x=data['A'], y=data['B'])

Python

Python

Python

with sns.axes_style('white'):
    sns.jointplot("A", "B", data, kind='kde');

Python

Python

Python

with sns.axes_style('white'):
    sns.jointplot("A", "B", data, kind="hex")

Pairplot

Let’s use the planets dataset from Seaborn Library for rendering multiple plots in a single image. Using pair plots will help in understanding paired relations among multiple variables.

planets = sns.load_dataset("planets")
planets.head()

Python

Python

Python

sns.pairplot(planets, hue='year', size=2.5);

Jointplot

In a joint plot, we get a bivariate graph and two separate graphs placed on their respective axis to show each variable’s distribution plotted with histograms.

In total, you get three graphs,

• one that shows the relationship between variables

• Other two represent variables individually.

Python

Python

Python

aplanets=sns.load_dataset('geyser')
sns.jointplot("duration", "waiting", data=planets, kind='reg')

kind parameter

We can change the kind parameter and use different arguments in the joint plot. To show an example, we will use the ‘hex’ argument with kind.

Python

Python

Python

geyser=sns.load_dataset('geyser')
with sns.axes_style('white'):
    sns.jointplot("duration", "waiting", data=geyser, kind='hex')

Factor Plot

For factor plots, we will use the taxis dataset, which has multivariate data and also has several data types. This dataset was originally published by the NYC Taxi and Limousine Commission (TLC)

taxis = sns.load_dataset('taxis')
taxis.head(5)

Python

Python

Python

with sns.axes_style(style='ticks'):
    g=sns.set(rc={'figure.figsize':(11.7,8.27)})
    g = sns.factorplot("color", "fare", "pickup_borough", data=taxis, kind="box")
    
    g.set_axis_labels("Taxis", "dropoff_borough");

Time Series Analysis

To understand time series analysis, we are using the worldwide life expectancy dataset. The first plot shows the world GDP over the years and its increase.

lifeexp=sns.load_dataset('healthexp')
lifeexp.head(1)

Python

Python

Python

with sns.axes_style('white'):
    g = sns.factorplot(x="Year",y="Spending_USD" ,data=lifeexp,aspect=4.0,)
    g.set_xticklabels(step=5)

Python

Python

Python

with sns.axes_style('white'):

    g = sns.factorplot(x="Year",y="Life_Expectancy" ,data=lifeexp ,aspect=2.0,)
    g.set_xticklabels(step=5)