Visual Encoding

What is data?

Data is an observation

a Phenomena

such as a recording of measurements

What are the types of data?

Nominal

Ordinal

Interval

Ratio

What is Nominal Data?

assigns measurement to a category, composed of categories.

EX: product categories

What is Ordinal data?

measurements in an order or sequence, defined by an order or sequence

EX: top 10 lists

What is interval data?

measurements in equal positions from each other, defined by an equal distance between units.

only relative values are significant, no absolute or zero point.

EX: temperature

What is ratio data?

measurements can be fractions / multiples, defined by numeric ratios of comparisons.

uses an absolute zero point

EX: money, hours, percentages compared to 0 or 100.

Continuous Data

Continuous data elements can take on any value between one number and another number (such as weight).

Discrete Data

Discrete data elements are counted as unique and separate units (such as number of servers, dog collars, and ice cream cones).

What is data representation?

A trend line represents a continuous measure

discrete measure defines individual columns

What is the graphics movement?

Jacques Bertin introduced the concept of graphical encoding.

Not artwork, it is a system of signs that convey meaning (IMPORTANT)

We will learn to see relationships in data

process of finding the right questions to answer.

What are the 8 variables of visual encoding?

X

Y

Size

Value

Texture

Color

Orientation

Shape

What are the variables of the image?

X

Y

Size

Value

What are the 2 dimensions of the plane?

X and Y

What are the two dimensions of Z?

Size and Value

What are the differential variables?

Texture

Color

Orientation

Shape

Value versus size

qualitatively compare the degree of black to white in 1. (slide 18)

value is represented as intensity

quantitatively compare the size of each box in 2.

Value encoding

value encoding can use a single color

the intensity of the color represents levels of a measure.

Texture encoding

Think of how manga panels and pages are colored with different screen tones, hatching, etc.

Color Encoding

red, green, and blue cones in our eyes receive color stimulus.

color is provided by filtering/reducing the amount of RGB

in RGB-A the “A” measures the amount of opacity

Order color light to dark to maximize power.

Orientation encoding

easier to interpret than shape

can be used in combination with size

Shape encoding

shapes must be a certain size to have visual meaning

when size contrast is low shapes have a small utility

Edward Tufte’s Principles

Data to ink ratio

reveal data at several levels of detail

make large data sets coherent (sparklines)

show the data - insights at many levels

induce viewer to think about the substance

integrate visualization with verbal definitions (annotations).

What is the data to ink ratio?

ink used on the page should have dense data encoding

create visually simple views

avoid adding a bunch of low value ink called chartjunk

Sparklines

Word + number + context of measurement

small graph that serves as a quick way to identify trends across multiple entries/ categories

Sparklines for baseball

sparklines provide context of measurements over time

small trend lines

What does show the data mean?

what is the summary mean of each x & y series (example)

Amscombe’s Quartet

at a granular level these data sets are very different

Creation of the matrix

Jacques invented the matric that we all know as the spreadsheet

Levels of information

Elementary

Intermediate

and Overall questions

Two types of questions (in regard to levels of information)

in X and in Y

Elementary question example

How many females visit the hotel in February

Processing Routine

every cell above mean average is filled in with black

every cell below the mean average is filled with white

sort all of the rows

Intermediate question example

relationship of the group of months, Dec, Jan, Feb.

To the grouping of characteristics on the x axis

Overall information level

generation of a set of insights

Statistical visualization consists of the

mean

median

and mode

what is the mean?

mathematical average of the value

What is the median?

middle measurement in order

what is the mode?

most frequent value or most common measurement.

What kind of distribution happens when the mean and median are the same?

symmetrical

what kind of distribution happens when the mean is greater than the median?

positively skewed (or right skew)

what kind of distribution happens when the mean is less than the median?

negatively skewed (or left skewed).

Box Plot

distribution analysis illustrates a distribution across a quantitative range

box plots are excellent for displaying multiple distributions.

Histogram

histograms displays a distribution breaking the data out by time segments or bins

number of patients are counted in each segment

Measurements in containers

box plots and histograms create bins of measurements

measurements are data (important)

Box plots vs histograms

histograms display the distribution as side by side bins

box plots represent the measurements in quartiles (25% sections).

Imputing values

it is common to have a field with missing data elements

if the field is valuable, we need to create the missing data values

we don’t just make up the data values

we impute the data values with the mean or median.

What to use when imputing for symmetric dists (mean or median?)

mean average

What to use when imputing for skewed dists (mean or median)?

median

Data visualization stack, from top to bottom

Visualization

Insights

These two feed into each other like a cycle.

Information

Data organization

Data source

Measurements

Phenomena

Think of the process of building a lego, you start from the beginning and build upwards (with tableau features)

Information vs data

information is a conceptual understanding or communication

EX: physical state of water changes at 0 degrees and 100 degrees.

Operates at a higher level than data (measurements)

EX of data: the actual degrees recorded.

What is an insight?

insights are new or fundamental understanding about the world

EX: matter has states of solid, liquid, and gas

Application or dynamics of understanding (communication) create viz

EX: physical state of water/elements changes at 0 degree and 100 degree.

Lecture summary

questions exist at multiple levels of information

primary variables of images are X and Y dimensions with size and value in the Z dimension

texture, color, orientation, and shape make up the differential variables

data is the capture of a phenomena

information is a synthesis of data measurements

insight is something unusual or significant