Data Visualization and Exploration

Progress report

Introduction

• Announcements
• Graph showcase
• Overview of software “ecosystems”
• Reflection on past, look to the future

Melton Gallery

Figure 1: Informational flyer for current Melton Gallery exhibit.

We will visit the gallery on April 1 for a special, interactive, behind-the-scenes gallery talk with the curator.

Graph showcase

This will not be an exhaustive display.

• Not all images from html files rendered.
• Some graphs are excerpted as examples from collections.

Sincere apologies if your graph is not represented today, we can try again later.

Disclaimer

These are exploratory graphs, likely at various stages of “completion”.

• You were not experts in the data, you were learning the software.
• You were exploring the data, an open-ended, creative process.
• You were still learning the software.

Please be gentle and constructive in your comments or questions.

Olympics data (medals)

Figure 2: Olympic medals as bar graph.

Olympics data (body metrics)

Figure 3: Body metrics of Olympic athletes.

Figure 4: Body metrics of Olympic athletes.

Olympics data (country, age)

Figure 5: Top medal counts by country.

Figure 6: Total medal counts by age.

Olympic data (by sport)

Figure 7: Participation by sport and country.

\(\text{CO}_2\) emissions data (1/3)

Figure 8: Emissions data.

\(\text{CO}_2\) emissions data (2/3)

Figure 9: Emissions data.

\(\text{CO}_2\) emissions data (3/3)

Figure 10: Emissions data.

CPI data (1/2)

Figure 11: “Corruption Perception Index”, not “Consumer Price Index”.

CPI data (2/2)

Figure 12: “CPI” for India.

Diabetes data

Figure 13: Diabetes.

Pharmaceutical spending (1/3)

Figure 14: Pharmaceutical spending.

Pharmaceutical spending (2/3)

Figure 15: Pharmaceutical spending.

Pharmaceutical spending (3/3)

Figure 16: Pharmaceutical spending.

GDP data I

Figure 17: GDP.

GDP data II

Figure 18: Nordic economy data.

Questions?

What questions do you have? Big or small?

Recap

Base R vs ggplot

Both approaches rely on “layers”.

Most importantly, both should lead to reproducible graphs.

Layering graphics

• with base R, use plot() followed by¹ any of lines(), points(), legend(), axis(), text(), mtext()
• with ggplot2, use ggplot(data = ...) combined using + with any sensible geom_type(), legend, axis, or annotation command

1. possibly proceeded by par().

ggplot and tidyverse

The + syntax in use of ggplot() can be read “followed by” or “and then”¹.

tidyverse is a collection of packages (i.e., “an ecosystem”), primarily for data manipulation.

Both are good libraries to be familiar with, as is base R.

1. In base R, the “layering” just happens as many commands are designed to add content to an existing plot.

tidyverse philosophy (excerpted)

The tidyverse has four guiding principles:

• It is human centered, i.e. the tidyverse is designed specifically to support the activities of a human data analyst.
• It is consistent, so that what you learn about one function or package can be applied to another, and the number of special cases that you need to remember is as small as possible.
• It is composable, allowing you to solve complex problems by breaking them down into small pieces, supporting a rapid cycle of exploratory iteration to find the best solution.
• It is inclusive, because the tidyverse is not just the collection of packages, but it is also the community of people who use them.

What’s next

Philosophical and Practical questions?

• How to decide which tool? Do you need the new one, the fancy one, the reliable and familiar one?
• How to choose which visualization?

Technical questions?

• Can we make maps? Yes.
• Can we make dashboards¹? Yes.

1. There seems to be an “anti-dashboard movement” at the moment (i.e., “are dashboards dead?” opinion pieces)

The pipe operator

Code written “in the tidyverse” relies heavily on the “pipe operator” or “piping”.

The good:

• The pipe, %>%, is analogous to the + that strings together layers in ggplot().
• The pipe may make code more readable as you pass output from one step to the next, thinking “and then”.

The “bad”:

• The pipe may make it hard to ensure that the correct output moves through the pipeline.
• I, personally, find this code much less readable.

Why %>%?

The pipe is meant to be an alternative to function composition.

• Without the pipe, commands are nested inside out - the last thing to be done is first.
• With the pipe, commands accumulate top to bottom - the last thing to be doe is last.

I like to create and heavily inspect temporary objects (e.g., table(), aggregate()) as I explore.

You might like the pipe operator, but are cautioned to ensure that it actually produces what you think it produced at each step.