5. Visualization

Dr. Andrew Abela created this chart chooser visualization in 2013. The four-dimension chart categorization offer great insight when choosing visualizations.

Fig. 5.1 Andrew Abela Chart Chooser

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

Overview

Data visualization is the art and science of transforming raw numbers into compelling visual stories. While spreadsheets and statistical summaries provide precise information, visualizations reveal patterns, relationships, and insights that would remain hidden in tables of data.

Imagine trying to understand weather patterns from thousands of temperature readings versus seeing them plotted as a line graph over time—the visual immediately shows trends, seasonal cycles, and anomalies that numbers alone cannot convey.

Each tool has its strengths: Python excels in programmatic control and integration with data analysis workflows, while R provides statistical visualization excellence, while Tableau and Power BI offer user-friendly interfaces for business users.

Why Visualization is Essential

Data visualization transforms numbers into understanding, serving three fundamental purposes:

• Exploration and Discovery: Visualization reveals patterns, outliers, and relationships invisible in raw data, guiding initial analysis and data cleaning decisions.

• Communication and Persuasion: Well-crafted visuals convey complex findings to diverse audiences, translating technical results into accessible insights.

• Decision Support and Action: Visualization provides clarity for confident decision-making through dashboards, trend analysis, and comparative displays.

Ultimately, visualization bridges the gap between data analysis and actionable insight.

Effective data visualization follows four key principles:

• Context is Key: Design for your specific audience and their decision-making needs.

• Keep It Simple: Use clear labels and remove unnecessary elements that don’t add value.

• Choose the Right Chart Type: Match chart types to data structure:

  • bars for categories,

  • lines for trends,

  • scatter plots for relationships.

• Tell a Story: Structure visualizations to guide the audience through a logical narrative flow.

Common Tools and Libraries

The data visualization landscape offers tools ranging from point-and-click platforms to code-based libraries. Understanding this ecosystem helps you choose the right tool for your needs.

Business Intelligence Platforms:

• Tableau: Industry-leading dashboard creation with drag-and-drop interface

• Power BI: Microsoft’s analytics platform with strong Excel and cloud integration

Python Visualization Libraries:

• Matplotlib: The foundational library providing complete control over plot elements

• Seaborn: Statistical graphics with simplified syntax, built on Matplotlib. Seaborn is a higher-level library specifically designed for creating more visually appealing and informative statistical graphics with less code. It’s excellent for exploring relationships between variables. Seaborn also requires Matplotlib as a dependency.

• Plotly: Interactive, web-ready visualizations including 3D and geographic plots

• Pandas: Quick plotting directly from DataFrames for exploratory analysis

Why Focus on Python?

While business tools excel at dashboard creation and user-friendly interfaces, Python libraries offer several advantages for data scientists:

• Integration: Seamless workflow from data analysis to visualization

• Reproducibility: Code-based plots can be version controlled and automated

• Customization: Complete control over every visual element

• Cost: Open-source tools reduce licensing expenses

This chapter focuses on Pandas visualization—the fastest way to create exploratory plots directly from your data.

Which Plot to Use?

It is essential to know which plots to use when the need for data visualization arises. Matplotlib (plt/ax), Pandas (df.plot), and Seaborn (.) each have a set of functions and methods for data visualization. You will learn about the functions and methods later, but here is a summary of them. Pay attention to what they are used for for now and use this table as a reference.

Plot Type	Pandas (df.plot)	Matplotlib (plt)	Seaborn (sns)	Uses
Line	.line()/.plot()	.plot()	.lineplot()	Time series, trends
Scatter	.scatter(x, y)	.scatter()	.scatterplot()	Correlations, relationships
Bar (vert.)	.bar()	.bar()	.barplot()	Category comparisons
Bar (horiz.)	.barh()	.barh()	.barplot(y, x)	Long category labels
Histogram	.hist()	.hist()	.histplot()	Distribution of numeric data
Box plot	.box()	.boxplot()	.boxplot()	Distribution + outliers
KDE/ Density	.kde()/ .density()	—	.kdeplot()	Smoothed distribution
ECDF	—	— .ecdf() (v3.8+)	.ecdfplot()	cumulative distribution
Area	.area()	.fill_between()	—	Cumulative trends
Stacked area	.area( stacked=True)	.stackplot()	—	Composition over time
Pie	.pie()(ser)	.pie()	—	Part-to-whole (use sparingly)
Hexbin	.hexbin(x, y)(df)	.hexbin()	.jointplot( kind="hex")(fig.)	Dense scatter (binning)
Violin	—	.violinplot()	.violinplot()	Distribution shape by category
Heatmap	—	.imshow()/ .pcolormesh()	.heatmap()	Correlation matrices, grids
Count plot	—	—	.countplot()	Frequency of categories
Pair plot	—	—	.pairplot()(fig.)	Multivariate relationships
Joint plot	—	—	.jointplot()(fig.)	Bivariate + marginals
Regression	—	—	.regplot()(ax)/ .lmplot()(fig.)	Linear relationships + CI
Contour	—	.contour()/ .contourf()	.kdeplot(x, y) (2D density)	Continuous 2D fields/level sets

Seaborn Datasets

Seaborn comes with several datasets that are commonly used for learning data science and machine learning.

Name	Rows × Cols	What it’s about	Common uses
tips	244 × 7	Restaurant bills & tips	categorical plots, regression, grouping
penguins	344 × 7	Palmer penguins (species & measurements)	scatter, hue, KDE, classification demos
iris	150 × 5	Iris flower measurements	pairplots, clustering, basics
diamonds	53,940 × 10	Diamond prices & attributes	regression, categorical + numeric
flights	144 × 3	Monthly air passengers (’49–’60)	heatmaps, time series
titanic	891 × 15	Titanic passengers	categorical analysis, missing data
planets	1,035 × 6	Exoplanet discoveries	distributions, facet grids
fmri	1,064 × 4	fMRI signal over time	lineplots with CIs
exercise	90 × 4	Exercise & pulse	catplots, faceting
anscombe	44 × 3	Anscombe’s quartet	scatter + regression; cautionary stats

To load the datasets, use the load_dataset function with syntax:

[name] = sns.load_dataset("[dataset]")

Here we would take a quick look at the popular ones.

Iris

The Iris dataset, introduced by Ronald A. Fisher in 1936, is one of the most well-known datasets in statistics and machine learning. It is often used for testing classification and visualization techniques. It contains measurements of three species of iris flowers: Setosa, Versicolor, and Virginica. The features of the iris dataset include:

Feature	Description	Units
sepal_length	Length of the outer part of the flower	cm
sepal_width	Width of the outer part	cm
petal_length	Length of the inner petal	cm
petal_width	Width of the inner petal	cm
species	Type of iris flower (setosa, versicolor, virginica)	categorical

### load dataset

iris = sns.load_dataset("iris")
iris.head(3)

	sepal_length	sepal_width	petal_length	petal_width	species
0	5.1	3.5	1.4	0.2	setosa
1	4.9	3.0	1.4	0.2	setosa
2	4.7	3.2	1.3	0.2	setosa

Tips

The tips dataset records restaurant bills and gratuities along with simple demographics, making it useful for practicing categorical comparisons and relationships between numeric variables.

The tips dataset has features such as:

Feature	Description	Type
total_bill	Total bill amount (including tax)	float
tip	Tip amount given by the customer	float
sex	Gender of the server’s customer (Male, Female)	categorical
smoker	Whether the party included smokers (Yes, No)	categorical
day	Day of the week (Thur, Fri, Sat, Sun)	categorical
time	Meal type (Lunch, Dinner)	categorical
size	Number of people in the dining party	integer

tips = sns.load_dataset('tips')
tips.head(3)

	total_bill	tip	sex	smoker	day	time	size
0	16.99	1.01	Female	No	Sun	Dinner	2
1	10.34	1.66	Male	No	Sun	Dinner	3
2	21.01	3.50	Male	No	Sun	Dinner	3

Titanic

The titanic dataset includes passenger demographics, ticket class, fares, and survival outcome, making it a classic dataset for classification and categorical analysis of the catastrophic event in 1912.

titanic = sns.load_dataset('titanic')
titanic.head(3)

	survived	pclass	sex	age	sibsp	parch	fare	embarked	class	who	adult_male	deck	embark_town	alive	alone
0	0	3	male	22.0	1	0	7.2500	S	Third	man	True	NaN	Southampton	no	False
1	1	1	female	38.0	1	0	71.2833	C	First	woman	False	C	Cherbourg	yes	False
2	1	3	female	26.0	0	0	7.9250	S	Third	woman	False	NaN	Southampton	yes	True

Planets

The planets dataset contains exoplanet discoveries, including discovery method, orbital period, and mass, and is useful for distributions and time-based summaries.

planets = sns.load_dataset('planets')
planets.head(3)

	method	number	orbital_period	mass	distance	year
0	Radial Velocity	1	269.300	7.10	77.40	2006
1	Radial Velocity	1	874.774	2.21	56.95	2008
2	Radial Velocity	1	763.000	2.60	19.84	2011

# Ensure the Titanic dataset is available for preview examples
if "sns" not in globals():
    import seaborn as sns

if "titanic" not in globals():
    titanic = sns.load_dataset("titanic")

Previewing Data

After you load a new dataset, always use the following methods/property to explore the data:

• head() (how the dataset looks like),

• describe() (descriptive statistics), and

• shape (get dimension information; or just evaluate the dataframe)

Here is a preview (head()) of the first five rows so you can see the raw values.

print("head(5):")
print(titanic.head(5))

head(5):
   survived  pclass     sex   age  sibsp  parch     fare embarked  class  \
0         0       3    male  22.0      1      0   7.2500        S  Third   
1         1       1  female  38.0      1      0  71.2833        C  First   
2         1       3  female  26.0      0      0   7.9250        S  Third   
3         1       1  female  35.0      1      0  53.1000        S  First   
4         0       3    male  35.0      0      0   8.0500        S  Third   

     who  adult_male deck  embark_town alive  alone  
0    man        True  NaN  Southampton    no  False  
1  woman       False    C    Cherbourg   yes  False  
2  woman       False  NaN  Southampton   yes   True  
3  woman       False    C  Southampton   yes  False  
4    man        True  NaN  Southampton    no   True  

Here is a quick summary of the numeric columns in the Titanic dataset.

print("describe():")
print(titanic.describe())

describe():
         survived      pclass         age       sibsp       parch        fare
count  891.000000  891.000000  714.000000  891.000000  891.000000  891.000000
mean     0.383838    2.308642   29.699118    0.523008    0.381594   32.204208
std      0.486592    0.836071   14.526497    1.102743    0.806057   49.693429
min      0.000000    1.000000    0.420000    0.000000    0.000000    0.000000
25%      0.000000    2.000000   20.125000    0.000000    0.000000    7.910400
50%      0.000000    3.000000   28.000000    0.000000    0.000000   14.454200
75%      1.000000    3.000000   38.000000    1.000000    0.000000   31.000000
max      1.000000    3.000000   80.000000    8.000000    6.000000  512.329200

Here is the dataset shape to confirm the number of rows and columns.

print("shape:", titanic.shape)

shape: (891, 15)

Style Sheets

The style sheet, or themes, basically creates a set of style rules that your plots follow. The use of a stylesheet gives your plots a unified look and feel, making them more professional. You can even create your own stylesheet.

Matplotlib has style sheets (or themes) you can use to make your plots look a little nicer. Popular stylesheets include:

• bmh (Bayesian Methods for Hackers)

• fivethirtyeight (FiveThirtyEight is a news site)

• ggplot (R’s ggplot2 default theme)

• dark_background

The syntax for using stylesheets in matplotlib is:

plt.style.use(style_name)

To see all the stylesheets available, use:

plt.style.available

Note that:

• we use plt.style, which means we are using matplotlib here.

• Pandas by default pulls colors from Matplotlib’s axes.prop_cycle, a Matplotlib rcParam (runtime configuration parameter), which is a color iterator (cycler) that cycles through a list of predefined colors. That’s why you may see different colors (by default, 3) when you plot multiple lines.

Before plt.style.use(), let’s draw this histogram (this should be the default look):

titanic['age'].hist()

<Axes: >

Call the stylesheet. Let’s use the ggplot theme:

plt.style.use('ggplot')

After applying plt.style.use(ggplot):

titanic['age'].hist()

<Axes: >

Now try plt.style.use(bmh):

plt.style.use('bmh')
titanic['age'].hist()

<Axes: >

fivethirtyeight

plt.style.use('fivethirtyeight')
titanic['age'].hist()

<Axes: >

A dark background theme:

plt.style.use('dark_background')
titanic['age'].hist()

<Axes: >

Let’s stick with the ggplot style for now.

plt.style.use('ggplot')