OMIS 473

delete

delete

Tableau data types 

1. text/ string/ character values

2. numeric/ integer values 

3. Date values

4. Date and time values 

5. Boolean (true/ false) values

6. Geographic values that are to be used with Maps

4 properties of every field 

1. Data type - how values are stored, string, number, date, boolean or geographic 

2. Role - how the field behaves - dimension or measure

3. Display - how it appears in the view - discrete or continuous 

4. Aggregation - how multiple measure values are summarized, SUM, AVG, COUNT, and more

Difference between dimensions vs measures

• Dimensions group or segment (ex: category, region, customer, order date)

• Measures are aggregated. Examples, sales, profit, quantity, discount

• Dimensions change the level of detail by creating groups of marks

• Decision rule: do i want to group by this field or calculate with it

Dimensions features 

• Usually categorical labels, dates, or identifiers 

• They organize data into groups and change the level of detail

• Ex: product, customer, region, order date

Measures features 

•  Numeric values that tableau aggregates

• Common aggregations: SUM, AVG, MIN, MAX, COUNT, COUNTD

  • Ex: sales, profit, quantity, discount

Discrete vs. continuous

• Discrete (blue): separate values that create headers 

• Continuous (green): a range that creates an axis

• Field role and display mode are independent decisions

delete

delete

Different variable types

• Nominal - categories have no natural order

• Ordinal - categories follow a meaningful order

• Discrete -counted values, typically separate whole numbers

• Continuous - measured values, any value within a range is possible 

Different bar charts (use to compare categories)

• Frequency bar chart

  • 1 categorical variable 

  • Shows frequency (count) of each category

• Aggregated bar chart

  • 1 categorical + 1 numerical 

  • Shows aggregated value (sum, mean, median) of numerical variable 

• Clustered bar chart

  • 2 categorical + 1 numerical 

  • 2 categorical + 2 numerical 

  • Compare values across groups of categories

• Stacked bar chart 

  • 2 categorical variables + 1 numerical variable 

  • Shows the composition of categories within each bar

Horizontal bar charts

• Useful when category names are long/ there are many categories or bars

• 1 dimension + 1 measure.

Clustered bar charts

• Aka Side by side bar chart, or group bar chart 

• Displays multiple bar charts side by side within each category

• How do subgroups compare within each main category?

• REQUIRED: 2 categorical + 1 numerical

Stacked bar chart

• When you want to show both the overall total and its composition

• Each complete bar represents the total for one category

• Each colored section shows how a subgroup contributes to that total

• Shows part to whole relationship

REQUIRED: 2 categorical + 1 numerical (not good for exact subgroup comparison)

When to use a stacked vs. clustered bar chart

• Use clustered when the main goal is to compare the exact values of individual subgroups across categories

When to use 100% stacked bar chart

when relative percentages are more important than total values

100% stacked bar chart -

displays every bar with the same total height of 100% 

• Each color represents the percentage contributed by one subgroup 

• Emphasizes relative composition 

• Good when main goal is to compare how subgroup proportions differ across categories

• REQUIRED: 2 categorical + 1 numerical converted to %

  • Does not show absolute totals

Calculated field

creates a new measure from existing variables

text table

Highlight table

Horizontal Bar chart frequency

Horizontal Bar chart aggregation

clustered bar

stacked bar

100% stacked bar

delete

delete

3 levels of memory

1. Iconic memory

   1. Starts from the sensory input 

   2. Pre-attentive attributes

   3. Unattended information is lost 

2. Short term memory 

   1. Maintenance rehearsal 

   2. Starts with attention and ends with encoding 

   3. Unrehearsed information is lost 

3. Long term memory

   1. Only develops after encoding 

Pie chart needs

one categorical dimension and one quantitative measure

Law of enclosure

individuals perceive objects that are physically enclosed together as belonging to the same group

Law of proximity

 Objects that are close to one another are perceived as a group

Law of Similarity

Elements within an assortment of objects are perceptually grouped together if they are similar to each other.

Law of continuity

when looking at objects, our eyes seek the smoothest path and naturally create continuity in what we see. Even if it might not actually exist 

Law of connection

people perceive objects that are physically connected as part of a group 

Scatter plot

 shows relationship between 2 numerical variables

• Each observation must contain a pair of numerical variables, an x value and y value. Each pair is represented by 1 point

• Can suggest a correlation between two variables 

• X (independent) explanatory variable

• Y (dependent) response variable 

• Correlation does not establish causation

Law of Closure (Law of Past Experience)

People mentally complete incomplete shapes and perceive them as whole object.

Decluttering data

1. Remove chart border

2. Remove gridline

3. Remove data markers

4. Clean up axis labels

5. Label data directly 

6. Leverage consistent color

Line charts

commonly used to plot continuous data

• Quantitative data on y axis and time on x axis

• Great for show trends over time

• Data must be plotted in consistent intervals

Histogram

distribution of one quantitative variable 

• Bins group observations into numerical intervals 

• Xaxis shows intervals in numeric order

• Y axis shows frequency percentage or density 

• 1 Quantitative measure; multiple observations

• Reads center, spread, shape, skew, frequency, bins

Box and whisker plot

Useful for detecting potential outliers and seeing spread and skewness of a data set

• REQUIRED: 1 measure + optional grouping dimension

• Median, Q1, Q3, IQR, whiskers, outliers

IQR

Q3 - Q1

Lower fence

 Q1 - 1.5 x IQR

Upper fence

Q3 + 1.5 x IQR

Whisker end points

most extreme observations inside the fences (within the box plot)

Strip plot

each dot represents an individual observation and clusters of a quantitative variable

• Separate strips compare distributions across categories

• Look for clusters, spread, gaps, and unusual values.

• REQUIRED: 1 measure + 1 observation ID; optional dimension

How to know which distribution chart to choose

• Histogram - look at shape of distribution 

• Boxplot - compare medians and spreads 

• Strip plot - look at individual observations

How fields change a scatter plot

two measures define position, a dimension determines the number and meaning of the marks 

Bubble scatter plot

keeps two position measures and adds a third measure through bubble area

• x measure + y measure + size measure

Slopegraph

pack in a lot of information, only showing two time periods or points of comparison

Q: which groups increased or decreased between two points

• group dimension + measure at two time points

Area charts

used to compare the overall trend + contribution of multiple data sets

• Plot continuous data over time with the area below the line filled in with color or shading to visually represent the magnitude of the data.

• Continuous data on the Y axis and the units of time on the X axis

• Great for showing the proportion of each data set relative to the whole

• Time + measure + subgroup + dimension

• Works for part to whole overtime

Box and Whisker plot components 

Main limitations of the charts

• Histogram - bin choices affect visible pattern

• Boxplot - clusters and exact values can be hidden

• Strip plot - dots can overlap

side by side circle chart

discrete line chart

slopegraph

area chart

Who/what/where/when prompt

usually a dimension or category

how much/ how many/ how long prompt

usually a quantitative measure

why/ how prompt

needs comparison, trend, relationship, or distribution

data source

Connect Tableau to Excel or another structured data file

• data source page; check field types

Rows / Columns

Place fields to define the chart structure

• Dimension → header; measure → axis

Marks card

Controls mark type, color, size, label, detail, tooltip

• change bar, pie, circle, line, area; add labels

Text table

Look up exact values by row/column.

• REQUIRED: Dimensions + measure(s).

Highlight table

Scan for high/low values quickly

• REQUIRED: Dimensions + measure encoded by color saturation.

Bar chart

Compare values across categories

REQUIRED: 1 categorical dimension + 1 measure.

• zero baseline needed

Basic bar

Which category has the highest value?

• REQUIRED: 1 categorical + 1 numerical

• Not for trends over time unless time is
  categorical.

calculated field bar

How does a derived metric compare?

• calculated measure + dimension

• Formula must match business meaning.

one meaningful whole (in pie chart) =

1 categorical dimension + 1 quantitative measure

Use a pie chart when

• goal is overall composition not exact ranking

• categories are mutually exclusive and collectively exhaustive

• few slices tell a simple story

• labels show percentages or absolute values clearly

use bar chart when

• there are many categories or small differences

• need precise comparison or ranking

• values include negatives or do not form a whole

• legend is long or hard to read

Filtered scatter

focus on relevant records or threshold cases

2 measures + filter condition

• reference lines or labels for threshold

Discrete line view

how do patterns compare across years/ quarters

• time parts as discrete headers + measure

• use when comparing period patterns

Distribution charts short summary

1. Histogram - (group values) changing bin width changes visual pattern

2. Box plot - (summarize) hide individual values and exact sample size

3. strip plot - (reveal individual records) use small/ transparent dots when many overlap

Composition charts

pie, stacked bar, 100% stacked bar, area

REQUIRED: Dimension + 1 measure

relationship charts

scatter/ bubble

REQUIRED: 2+ measures/ optional dimension

trend over time charts

line, slopegraph, area

REQUIRED: time + measure; optional dimension

distribution charts

histogram, box plot, strip plot

1 measure + many observations

nominal/ unordered variables

need distinct hues

• ex: product category, region, class, purpose

ordered/ quantitative variables

need sequential light to dark ordering

• ex: low to high sales, count, rate

deviation around a meaningful midpoint

means diverging a palette centered at 0 or target

• ex: profit/ loss, above/ below benchmark

Strip plot visual