Data Visualization

Lecture 5: Data Viz II

Author

Robert W. Walker

Outline

  • Readings: Chapters 1 and 2 and Wickham

Some Data

Google Sheets

A Data Taxonomy

  • Generally column-centric.
  • Variables in columns.
  • Units in rows.

Long data

Long data

This is how Python also thinks about long data.

Chapter 2

  • Means and standard deviations [TBC]
  • Boxplots and percentile statistics [TBC]
  • Histograms, densities, and shape
  • Outliers, robust statistics, and shape
  • Transformations

A note on context, or why you want to upload the csv [or similar]

Summarizing Data

Summary Statistics

Prompt

Prompt

Output

Output

Output

Seaborn is the current tool…

All kinda of possibilities

Seaborn-Cat

Start with a categorical variable

What can we do? Generically, nothing more than frequency. A table or a figure.

Table

Tables

Tables

Bar plot

Bar plot

Bar plot

A Lollipop Chart

could you turn that into a lollipop chart

lollipop

lollipop

Two Categorical Variables [Stacked]

Prompt and Result

Prompt and Result

Two Categorical Variables [Dodge]

Dodgy

Dodgy

A Mosaic

Could you constuct a mosaic plot of risk and fees?

Quantitative Variables

The core idea is they exist on a continuum; the likelihood of any particular realization is usually zero.

As a result:

  • we bin them.
  • we smooth their incidence.

Quantitative Variables

Show a histogram of assets

Controlling bins

show a histogram of assets with bins of size 100 million starting with 0 to 100 million

Densities

Boxplots

Qualitative/Quantitative Combinations

A Combination of Three Plots

A violin, boxplot, and raincloud

The boxenplot

How’s that done? 
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt


# Create the boxenplot for 'Return 2009' by 'Type'
plt.figure(figsize=(8, 5))
sns.boxenplot(x='Type', y='Return 2009', data=df, palette='muted')

# Styling
plt.title('Boxenplot of 2009 Returns by Fund Type')
plt.xlabel('Fund Type')
plt.ylabel('Return 2009 (%)')
plt.grid(axis='y', linestyle='--', alpha=0.3)

plt.tight_layout()

Violins

Note

Prompt: Can you show a violin plot of returns in 2009 by risk and type?

How’s that done? 
# Set categorical order for Risk
risk_order = ['Below average', 'Average', 'Above average']
df['Risk'] = pd.Categorical(df['Risk'], categories=risk_order, ordered=True)

# Create the violin plot
plt.figure(figsize=(8, 5))
sns.violinplot(x='Risk', y='Return 2009', hue='Type', data=df, split=True, palette='muted', inner='quartile')

# Styling
plt.title('Violin Plot of 2009 Returns by Risk and Fund Type', fontsize=14)
plt.xlabel('Risk Level', fontsize=12)
plt.ylabel('Return 2009 (%)', fontsize=12)
plt.legend(title='Fund Type')
plt.grid(axis='y', linestyle='--', alpha=0.3)

plt.tight_layout()

Adjustment

Note

Prompt: That is quite nice. I would prefer that they be oriented horizontally with each rising above the horizontal axis

Horizontals

Horizontals

Bandwidth adjustment

Final

Final

Two Quantitative Variables

Scatterplots [Lines just connect points]

Important

Prompt: create a scatterplot of returns by assets

Prompt: create a scatterplot of expense ratios and returns with symbols from fees and colors from risk

Important

Prompt: use a checkmark for yes and an X for no as the symbols

plotly

A Class Assignment to Conclude