Seaborn is a Python data visualization library based on Matplotlib. In this project, I explore Seaborn. I discuss Seaborn API overview, its functionality, setting Seaborn aesthetic parameters and colour palette. I discuss different distributions, various plot types and multi-plot grids with seaborn.
===============================================================================
The contents of this project is divided into various sections which are listed below:-
-
Introduction to Seaborn
-
Comparison of Seaborn and Matplotlib
-
Seaborn API Overview
-
Seaborn functionality
-
Seaborn installation
-
Import Python libraries
-
Import Seaborn
-
Import datasets
-
Set aesthetic parameters with set() method
-
Seaborn colour palette
-
Plotting Univariate Distribution with distplot() function
-
Seaborn – Histogram
-
Seaborn – Kernel Density Estimation
-
Plotting Bivariate Distribution with jointplot() function
-
Seaborn – Scatter Plot
-
Visualizing pairwise relationship with pairplot() function
-
Plotting categorical data
-
Seaborn – Strip Plot
-
Seaborn – Swarm Plot
-
Distribution of Observations
-
Seaborn – Box Plot
-
Seaborn – Violin Plot
-
Statistical estimation with Seaborn
-
Seaborn – Bar Plot
-
Seaborn - Point Plot
-
Linear relationships with Seaborn
-
Seaborn – Reg plot
-
Seaborn – Lm plot
-
Seaborn – Resid plot
-
Matrix plots with Seaborn
-
Seaborn – Heat map
-
Seaborn – Cluster map
-
Multi-plot grids with Seaborn
-
Seaborn – Facet Grid
-
Seaborn – Pair Grid
-
Seaborn – Joint Grid
-
Summary
-
References
===============================================================================
Seaborn is a data visualization library in Python. It is used to produce statistical graphics in Python. It is built on top of Matplotlib. It also supports NumPy and Pandas data structures. It also supports statistical units from SciPy.
In the world of analytics, data visualization plays an important role to explore and understand the data. Seaborn is developed to make this process of exploring and understanding the data smooth. It provides a high-level interface to produce high quality statistical graphics. The customizability and backend options for Matplotlib makes it easy to generate publication-quality figures.
In this project, I discuss Seaborn and present a high level overview of various plotting functions and tools associated with Seaborn.
===============================================================================
Seaborn uses Matplotlib to draw plots. Many actions can be accomplished with only Seaborn functions. Further customization options might require using Matplotlib directly. Seaborn is not an alternative to Matplotlib. We can think of it as a complement to Matplotlib. As it is built on top of Matplotlib, we can call Matplotlib functions directly for creating simple plots.
It is summarized that –
“If Matplotlib tries to make easy things easy and hard things possible, Seaborn tries to make a well-defined set of hard things easy too in a well-defined way.”
Seaborn helps to resolve two major problems faced by Matplotlib. The problems are:-
-
Working with Matplotlib’s default parameters
-
Working with data frames
===============================================================================
Seaborn has various types of plots which can satisfy various types of data visualization requirements. The plots can be categorized into seven categories – relational plots, categorical plots, distribution plots, regression plots, matrix plots and multi-plot grids.
1. Relational plots
We can visualize statistical relationships with relational plots. There are three types of relational plots – relplot(), scatterplot() and lineplot(). We can use these plotting functions to see trends and patterns that indicate a relationship.
2. Categorical plots
If one of the variable is categorical, we can visualize the relationship involving categorical data. In Seaborn, there are several different ways to visualize a relationship involving categorical data. There are various functions for plotting categorical data. We can plot categorical scatterplots with stripplot() and swarmplot() functions. Similarly, we can plot categorical distribution plots with boxplot(), violinplot() and boxenplot() functions. Also, we can plot categorical estimate plots with pointplot(), barplot() and countplot() functions.
3. Distribution plots
We can visualize the distribution of a dataset with the distribution plotting functions. We can plot univariate distribution with distplot() function. It will draw a histogram and fit a kernel density estimate (KDE). The kdeplot() function is used for plotting the shape of a distribution. We can use rugplot() function to plot data points in an array as sticks on an axis. We can draw a plot of two variables with bivariate and univariate graphs with jointplot() function. We can visualize pairwise relationships in a dataset using pairplot() function.
4. Regression plots
In Seaborn, there are various functions for visualizing linear relationships in a dataset. There are two main functions that are used to visualize linear relationship between variables. These functions are regplot() and lmplot(). They are closely related and share much of their core functionality. There is a third function called residplot() which is used to plot the residuals of a linear regression model.
5. Matrix plots
There are two plotting functions which are used to plot matrix plot types. These are heatmap() and clustermap(). Heatmap() is used to plot rectangular data as a colour-encoded matrix. Clustermap() is used to plot a matrix dataset as a hierarchically-clustered heatmap.
6. Multi-plot grids
While exploring medium-dimensional data, a useful technique is to draw multiple instances of the same plot on different subsets of the dataset. It allows us to extract large amount of information about complex data. Seaborn provides plotting functions to link the structure of the plot to the structure of the dataset and draw multi-plot grids.
To use these features, the dataset has to be in a Pandas dataframe. It must be in the tidy data format. It means that the dataframe should be structured such that each column is a variable and each row is an observation.
Multi-plot grids are further divided into three categories - Facet grids, Pair grids and Joint grids.
Facet grids provide three functions. FacetGrid(), FacetGrid.map() and FacetGrid.map_dataframe(). These functions are used to visualize the distribution of a variable or the relationship between multiple variables separately within subsets of the dataset.
Pair grids provide plotting functions to visualize pairwise data relationships. The basic plotting function is PairGrid(). It allows us to draw a grid of small subplots using the same plot type to visualize data. In a PairGrid(), each row and column is assigned to a different variable. So the resulting plot shows each pairwise relationship in the dataset.
The third category of plot is Joint grids. It provide plotting functions to draw Joint plots. There are four types of Joint grid plotting functions. They are JointGrid(), JointGrid.plot(), JointGrid.plot_joint() and JointGrid.plot_marginals(). JointGrid() function is used to draw a bivariate plot with marginal univariate plots. JointGrid.plot() is a shortcut to draw the full plot. JointGrid.plot_joint() is used to draw a bivariate plot of x and y. JointGrid.plot_marginals() is used to draw univariate plots for x and y separately.
===============================================================================
Seaborn offers lot of functionality which makes it effective for many data visualization tasks. These functionalities are listed below:-
• Seaborn provides a dataset-oriented API to examine relationships between variables.
• It provides functions to fit and visualize linear regression models for different types of independent and dependent variables.
• Seaborn provide functions for visualizing univariate and bivariate distributions and for comparing them between subsets of data.
• It provides plotting functions for using categorical variables to show observations or aggregate statistics.
• It helps us to visualize matrices of data and use clustering algorithms to discover structure in those matrices.
• Seaborn provides a plotting function to plot statistical time series data. The function provides flexible estimation and representation of uncertainty around the estimate.
• It provide tools for choosing styles, colour palettes, palette widgets and utility functions. These tools help us to make beautiful plots that reveal patterns in the data.
• It provides several built-in themes for producing stylish looking Matplotlib graphics.
===============================================================================
To install the latest release of Seaborn, we can use pip and run the following command in the terminal:
pip install seaborn
It is also possible to install the released version using conda as follows:
conda install seaborn
Alternatively, we can use pip to install the development version directly from github:
pip install git+https://github.com/mwaskom/seaborn.git
Dependencies
• Seaborn requires Python 2.7 or 3.5+ platform.
Mandatory dependencies
Seaborn requires following mandatory Python packages to be installed on the system.
• NumPy(>= 1.9.3)
• Scipy (>= 0.14.0)
• Matplotlib (>= 1.4.3)
• Pandas (>= 0.15.2)
Recommended dependencies
Seaborn also needs statsmodels (>= 0.5.0) package to be installed on the system.
===============================================================================
In this section, we import the necessary Python libraries Numpy, Pandas and Matplotlib with their usual shorthand notation.
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline
===============================================================================
We can import Seaborn as
import seaborn
The shorthand notation of Seaborn is sns. So, we can import seaborn with usual shorthand notation.
import seaborn as sns
===============================================================================
Seaborn has few important datasets in the library. When Seaborn is installed, the datasets download automatically. We can download any dataset with load_dataset() function.
Importing data as Pandas dataframe
In this section, we will import a dataset. By default, this dataset loads as Pandas dataframe. There is a dataset called tips in the library. We can import this dataset with the following line of code:-
tips = sn.load_dataset(‘tips’)
We can view the first five lines of code with the following command:-
print(tips.head())
To view all the available datasets in the Seaborn library, we can use the following command with the get_dataset_names() function as shown below:-
print(sns.get_dataset_names())
===============================================================================
We can set the aesthetic parameters of Seaborn plots with the set() method. The aesthetic parameters are context, style, palette, font, font_scale, color_codes, dictionary of rc parameters.
So, we can set the default aesthetic parameters by calling Seaborn's set() method.
sns.set()
===============================================================================
Colour plays very important role in data visualization. Colour adds various dimensions to a plot when used effectively. A palette means a flat surface on which a painter mixes paints.
Seaborn provides a function called color_palette(). It can be used to give colours to plots and adding aesthetic value to it. It return a list of colors defining a color palette.
There are several readily available Seaborn palettes. These are:-
• Deep
• Muted
• Bright
• Pastel
• Dark
• Colorblind
Besides these we can also create new palettes.
There is another function seaborn.palplot() which deals with color palettes. This function plots the color palette as a horizontal array.
Qualitative Color Palettes
Qualitative or categorical palettes are best suitable to plot the categorical data as follows:-
current_palette1 = sns.color_palette()
sns.palplot(current_palette1)
plt.show()
We can see the desired number of colors by passing a value to the n_colors parameter. Here, the palplot() function is used to plot the array of colors horizontally.
Sequential Color Palettes
Sequential plots are suitable to express the distribution of data ranging from relative lower values to higher values within a range. Appending an additional character “s” to the color passed to the color parameter will plot the Sequential plot.
We need to append ‘s’ to the parameter like ‘Greens’ as follows:-
current_palette2 = sns.color_palette()
sns.palplot( sns.color_palette(“Greens”))
plt.show()
Diverging Color Palette
Diverging palettes use two different colors. Each color represents variation in the value ranging from a common point in either direction. We assume plotting the data ranging from -1 to 1. The values from -1 to 0 takes one color and 0 to +1 takes another color.
By default, the values are centered from zero. We can control it with parameter center by passing a value as follows:-
current_palette3 = sns.color_palette()
sns.palplot( sns.color_palette(“BrBG”, 7))
plt.show()
Default Color Palette
We can set the default color palette of a Seaborn plot using set_palette() function. The arguments are same for both set_palette() and color_palette() functions, but the default Matplotlib parameters are changed so that the palette is used for all plots.
def sinplot(flip=1):
x = np.linspace(0, 15, 100)
for i in range(1, 5):
plt.plot(x, np.sin(x + i * .5) * (7 - i) * flip)
sns.set_style("white")
sns.set_palette("husl")
sinplot()
plt.show()
===============================================================================
The most important thing to do while analysing the data is to understand its distribution. Seaborn helps us to understand the univariate distribution of data.
The distplot() function provides a quick look at univariate distribution. This function will plot a histogram that fits the kernel density estimate of the data.
We can use the distplot() function as follows:-
sns.distplot(tips['total_bill'])
plt.show()
===============================================================================
Histograms represent the data distribution by forming bins along the range of the data and then drawing bars to show the number of observations that fall in each bin.
We can use the same distplot() function to plot a histogram as follows:-
sns.distplot(tips['total_bill'], kde=False)
plt.show()
The kde parameter is set to false. As a result, the representation of the kernel estimation plot will be removed and only histogram is plotted.
===============================================================================
Kernel Density Estimation (KDE) is a way to estimate the probability density function of a continuous random variable. It is used for non-parametric analysis.
Setting the hist parameter to false in distplot() function will yield the kernel density estimation plot.
sns.distplot(tips['total_bill'], hist=False)
plt.show()
===============================================================================
Bivariate Distribution is used to determine the relation between two variables. This mainly deals with relationship between two variables and how one variable is behaving with respect to the other.
The best way to analyze Bivariate Distribution in seaborn is by using the jointplot() function.
Jointplot() creates a multi-panel figure that projects the bivariate relationship between two variables and also the univariate distribution of each variable on separate axes.
sns.jointplot(x="total_bill", y="tip", data=tips)
plt.show()
===============================================================================
A scatter plot can be used to demonstrate relationship between two variables x and y. A simple scatter plot can be drawn as follows:-
sns.scatterplot(x="total_bill", y="tip", data=tips)
plt.show()
The relationship between the variables can be shown for different subsets of the data using the hue, size and style parameters.
===============================================================================
Some datasets contain many variables. In such cases, the relationship between each and every variable should be analysed. So, we need to plot pairwise relationships in a dataset.
To plot multiple pairwise bivariate distributions, we can use pairplot() function. This shows the relationship for (n,2) combination of variable in a dataframe as a matrix of plots and the diagonal plots are the univariate plots.
We can plot a pairplot as as follows:-
sns.set_style("ticks")
sns.pairplot(tips)
===============================================================================
So far, I have covered histogram, scatter plot and kde plots. They are used to analyze the continuous variables under study. These plots are not suitable when the variable under study is categorical.
When one or both the variables under study are categorical, we can use plots like striplot() and swarmplot() to plot categorical data.
===============================================================================
A stripplot() is used to draw a scatterplot where one variable is categorical. It represents the data in sorted order along any one of the axis.
We can plot a stripplot as follows:-
sns.stripplot(x="day", y="total_bill", data=tips)
plt.show()
We can add jitter to bring out the distribution of values as follows:-
sns.stripplot(x="day", y="total_bill", data=tips, jitter=True)
plt.show()
===============================================================================
Another option which can be used as an alternate to ‘jitter’ is to use function swarmplot(). This function positions each point of scatter plot on the categorical axis and thereby avoids overlapping points.
So swarmplot() can be used to draw categorical scatterplot with non-overlapping points.
We can plot a swarmplot as follows:-
sns.swarmplot(x="day", y="total_bill", data=tips)
plt.show()
===============================================================================
When dealing with a dataset, the first thing that we want to do is get a sense for how the variables are distributed.
We can use box plots and violin plots to plot the data distribution.
===============================================================================
We can draw a box plot to show distributions with respect to categories. Boxplot is a convenient way to visualize the distribution of data through their quartiles.
Box plots usually have vertical lines extending from the boxes which are termed as whiskers. These whiskers indicate variability outside the upper and lower quartiles, hence Box Plots are also termed as box-and-whisker plot. Any Outliers in the data are plotted as individual points.
We can plot a box-plot as follows:-
sns.boxplot(x=tips["total_bill"])
plt.show()
We can draw a vertical boxplot grouped by a categorical variable as follows:-
sns.boxplot(x="day", y="total_bill", data=tips)
plt.show()
===============================================================================
Violin Plots are combinations of the box plot with the kernel density estimates. So, these plots are easier to analyze and understand the distribution of the data.
We can draw a single horizontal violin plot as follows:-
sns.violinplot(x=tips["total_bill"])
plt.show()
We can draw a vertical violinplot grouped by a categorical variable as follows:-
sns.violinplot(x="day", y="total_bill", data=tips)
plt.show()
===============================================================================
Sometimes, we need to estimate central tendency of a distribution. Mean and median are the very often used techniques to estimate the central tendency of the distribution.
We can use bar plot and point plot to measure the central tendency of a distribution.
===============================================================================
A barplot shows the relation between a categorical variable and a continuous variable. The data is represented in rectangular bars where the length of the bar represents the proportion of the data in that category.
A barplot show point estimates and confidence intervals as rectangular bars. So, it represents the estimate of central tendency.
We can draw a set of vertical bar plot grouped by a categorical variable as follows:-
sns.barplot(x="day", y="total_bill", data=tips)
plt.show()
===============================================================================
Point plots serve same as bar plots but in a different style. Rather than the full bar, the value of the estimate is represented by the point at a certain height on the other axis. The Point plot show point estimates and confidence intervals using scatter plot glyphs.
We can draw a set of vertical point plots grouped by a categorical variable as follows:-
sns.pointplot(x="time", y="total_bill", data=tips)
plt.show()
===============================================================================
Many datasets contain multiple quantitative variables, and the goal of an analysis is often to relate those variables to each other. We can use statistical models to estimate a simple relationship between two sets of observations. These are termed as regression models.
While building the regression models, we often check for multicollinearity, where we had to see the correlation between all the combinations of continuous variables and will take necessary action to remove multicollinearity if exists.
There are two main functions in Seaborn to visualize a linear relationship determined through regression. These functions are regplot() and lmplot(). There is a third function residplot() that plot the residuals of a linear regression model.
===============================================================================
The function regplot() plot data and a linear regression model fit. It draws a scatterplot of two variables, x and y, and then fit the regression model y ~ x and plot the resulting regression line and a 95% confidence interval for that regression.
We can draw the scatterplot and regression line using regplot() as follows:-
sns.regplot(x="total_bill", y="tip", data=tips)
plt.show()
===============================================================================
The function lmplot() plot data and regression model fits across a FacetGrid. This function combines regplot() and FacetGrid. It is intended as a convenient interface to fit regression models across conditional subsets of a dataset.
We can plot a simple linear relationship between two variables using lmplot() as follows:-
sns.lmplot(x="total_bill", y="tip", data=tips)
plt.show()
The regplot() and lmplot() functions are closely related, but the former is an axes-level function while the latter is a figure-level function that combines regplot() and FacetGrid.
===============================================================================
The function residplot() plot the residuals of a linear regression. We can plot the residuals as follows:-
sns.residplot(x="total_bill", y="tip", data=tips)
plt.show()
===============================================================================
There are two functions which enable us to plot data in the form of a matrix. These are heat map and cluster map.
===============================================================================
Seaborn heatmap() function plot rectangular data as a color-encoded matrix.
We can plot a heatmap for a numpy array as follows:-
uniform_data = np.random.rand(10, 12)
sns.heatmap(uniform_data)
plt.show()
===============================================================================
Seaborn clustermap() function plot a matrix dataset as a hierarchically-clustered heatmap.
We can plot a clustered heatmap of tips dataset as follows:-
df1 = tips[['total_bill', 'tip', 'size']]
sns.clustermap(df1)
plt.show()
===============================================================================
When exploring medium-dimensional data, a useful approach is to draw multiple instances of the same plot on different subsets of the dataset. This technique is sometimes called either “lattice”, or “trellis” plotting, and it is related to the idea of drawing multiple instances of the same plot. It allows us to quickly extract a large amount of information about complex data.
Seaborn provides three types of grids – facet grid, pair grid and joint grid to draw multiple instances of the same plot.
===============================================================================
Seaborn Facetgrid() function enable us to draw multi-plot grid for plotting conditional relationships.
First of all, we initialize a 2x2 grid of facets using the tips dataset as follows:-
sns.FacetGrid(tips, col="time", row="smoker")
plt.show()
Initializing the grid like this sets up the matplotlib figure and axes, but doesn’t draw anything on them.
The main approach for visualizing data on this grid is with the FacetGrid.map() method. We need to provide it with a plotting function and the name(s) of variable(s) in the dataframe to plot.
We can draw a univariate plot on each facet as follows:-
g = sns.FacetGrid(tips, col="time", row="smoker")
g = g.map(plt.hist, "total_bill")
plt.show()
===============================================================================
Seaborn PairGrid() function draws subplot grid for plotting pairwise relationships in a dataset.
We can draw a scatterplot for each pairwise relationship using PairGrid() function as follows:-
g = sns.PairGrid(tips)
g = g.map(plt.scatter)
plt.show()
We can show a univariate distribution on the diagonal as follows:-
g = sns.PairGrid(tips)
g = g.map_diag(plt.hist)
g = g.map_offdiag(plt.scatter)
===============================================================================
Seaborn JointGrid() function help us to set up the grid of subplots.
We can initialize the figure but don’t draw any plots onto it as follows:-
g = sns.JointGrid(x="total_bill", y="tip", data=tips)
We can add plots using default parameters as follows:-
g = sns.JointGrid(x="total_bill", y="tip", data=tips)
g = g.plot(sns.regplot, sns.distplot)
===============================================================================
In this project, I present a high level overview of Seaborn and associated plots.
I discuss Seaborn and compare it with Matplotlib. I present Seaborn API overview and discuss Seaborn functionality. I discuss how to set aesthetic parameters of a plot and how to set color styles.
Then, I discuss various types Seaborn plots. I plot univariate distribution with distplot() function. Then, I discuss histogram and kernel density estimation plots. I plot bivariate distribution with jointplot() function and discuss Seaborn scatter plot.
I visualize pairwise relationship with pairplot() function. Then, I plot categorical data with Seaborn strip plot and swarm plot. I visualize the distribution of observations with Seaborn box plot and violin plot. I measure the statistical estimates with Seaborn bar plot and point plot.
I visualize the linear relationships between variables with Seaborn reg plot and lm plot. I visualize the residuals with Seaborn resid plot. I discuss Seaborn heat map and cluster map. Then, I discuss multi-plot grids with Seaborn. I discuss facet grid, pair grid and joint grid.
===============================================================================
The Seaborn plots and galleries in this project are taken from the following websites:-
-
Seaborn official documentation (https://seaborn.pydata.org/)
-
https://jakevdp.github.io/PythonDataScienceHandbook/04.14-visualization-with-seaborn.html
-
Seaborn tutorial from tutorialspoint