100% found this document useful (1 vote)
118 views

Data Storytelling Notes

This document provides an overview of principles and best practices for data storytelling. It discusses using narrative, visuals and data to effectively communicate insights. Specific chart types like bar charts, histograms, line charts and scatter plots are examined in terms of their best uses. Guidance is given for selecting visual variables, using color and designing for pre-attentive attributes. Storytelling frameworks like SPSN and Minto's pyramid are introduced. The importance of structure, scope and considering the audience are emphasized for effective data storytelling presentations.

Uploaded by

Keng Whye Leong
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
100% found this document useful (1 vote)
118 views

Data Storytelling Notes

This document provides an overview of principles and best practices for data storytelling. It discusses using narrative, visuals and data to effectively communicate insights. Specific chart types like bar charts, histograms, line charts and scatter plots are examined in terms of their best uses. Guidance is given for selecting visual variables, using color and designing for pre-attentive attributes. Storytelling frameworks like SPSN and Minto's pyramid are introduced. The importance of structure, scope and considering the audience are emphasized for effective data storytelling presentations.

Uploaded by

Keng Whye Leong
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 20

Module 1.

1 - Fundamentals of Data Storytelling


- Principles of Good Data Storytelling
● Narrative
● Visuals
- Purpose is to reduce time to insights
- Increase accuracy of insights
- Improve engagement
● Data

- Best practices
● Remove clutter
● Use color purposefully & sparingly
● Add the right visual elements to communicate insights
● Use a ‘storytelling chart title’ instead of just naming the chart as is

- Example 1: Bad Example vs Good Example

● Months are horizontal -> Not good, audience need to tilt head
● Usage of bar chart does not really highlight the intended message
● Title is merely descriptive, does not tell the story
● Decimal points on y-axis is redundant and adds unnecessary clutter

- Types of Data
● Numerical
(1) Discrete
- Whole number
- E.g. Number of children, number of pencils
(2) Continuous
- Data that can take decimal points
- E.g. Height, Weight
● Categorical
(1) Nominal
- E.g. Gender, blood type, country
(2) Ordinal
- Categorical data that is ordered
- E.g. Ranks in a company

- Bar Chart / Histograms


● Histogram v Bar Chart
- Histogram is a subset of bar chart
- Histogram is used to show a range of numeric values & show data summary
- Bar chart is used to display categorical data & show comparisons between categories

Histogram

- Histogram gives you a sense


of the distribution of the data
- Histogram also has a
summary effect cause you’re
effectively reducing number of
data points when you bin it
- General idea is that you want
to control the histogram’s bin
size to show the
story/trend/pattern you might
want to highlight

Bar Chart

- Best Practices
● No vertical/slanted texts on the axis labels
● Font-size on chart texts should be sized reasonably
● Sort/order the bars if possible
● Try to label directly on the bar chart instead using a legend
● Remove/lighten unnecessary grids/borders
● Use ‘Others’ to consolidate categories if there are too many

Other Forms of Bar Chart - Column Chart (Two types)

- Note the advantages/disadvantages between these two column bar charts


● Clustered - You can do a category to category comparison across months
● Stacked - You can see the proportion between categories within a month

- Line Chart
● Useful for examination of data over a period of time; usually you put time on the x-axis
● You can see the peaks/dips of the data over a time period -> Some questions:
- What is happening?
- Why is this happening?
- Is this seasonal?
● Best Practices
(1) Don’t use line chart for comparison over categorical data; use it to illustrate trends

(2) Avoid spaghetti graph -> Too many lines on a single chart

- One of the ways to avoid spaghetti graphs is to


use ‘contrast’; highlight what is important and dim
those that are not so important
(3) Avoid using legend; directly label the lines

(4) Remove grid lines and y-axis if possible

- Pie Chart
● Pie Chart is bad -> Hard to tell relative size of bars
- Instructor: Where possible, PLEASE DO NOT USE A PIE CHART

● Best Practices
(1) Do not use a 3D pie chart
(2) Do not use legend; direct label the slices
(3) Start at 12 o’clock, then go clockwise direction

(4) Avoid too many segments


(5) Use shades of same color to represent proportion

● Alternatives to Pie Chart


- Use a bar to indicate proportion

- Combo Charts: Dual Axis Chart

- Waterfall Chart
● Good to show changes from a starting point to ending point
● Color used well to illustrate increase/decrease

- Scatter Plot

Bubble Plots - Additional dimension to Scatter Plot

- Symmetry Charts
- Suggestions on how to Select a Chart
Module 1.2 - Principles of Data Visualization
- Pre-attentive Attributes
● Refers to things that your brain subconsciously process first when you look at something visually; I.e. Things that you
see and process first visually
● Make use of this concept to allow your audience to quickly communicate information visually

- Human brain will tend to see generalize patterns in things


and gloss over odd stuff/details
- Use contrast/dimming/colors to bring out the data story

- See right for textual highlights

- Usage of colors
● In general, avoid using colors to beautify a chart
● Using colors with established conventions/connotations
- E.g. Red = Bad, Green = Good

● Using colors to create visual hierarchies

- Visual Variables
● Depending on the type of variable that you have, there is hierarchy in how accurate it can be visually depending on
the shape/dimension
- Gestalt Principles
● Spatial arrangement can affect perception
● Idea - You wanna subtract useless things as far as possible. Beauty is simplicity. Less things to process

● Color of foreground/background can direct the audience to different messages/stories

● Closure

● Continuity
● Symmetry

● Similarity

- Graphical Integrity
● Truncated axis can make a trend appear more/less steep
● Chart on the right make the increase seem steeper than it really is because y-axis scale starts at 94m
● If you really need to truncate the axis, you can declare or make it obvious to the audience

- Juxtaposition between (A) Just


showing data (B) Storytelling with Data
● Concept of annotating the chart with relevant information to tell a story
Module 2.1 - Tableau Workshop
Module 3.1 - Data Storytelling Narratives
- Narrative Structure Framework:
(1) SPSN
● Storytelling dimension to this -> Compare this to Minto pyramid

● Example:

(2) Minto Pyramid


● Generally used by consultants
● Generally preferred by audience who are more ‘objective’
● Try to keep number of points to at most 3; ‘Human minds have limited capacity’

● Example

(3) GQM - Goal Question Metric

● Goal: Defined for specific purpose, w focus on deriving meaningful outcomes for business
- E.g. Improve model performance
● Question: Set of questions used to assess goal achievements
● Metric: Set of metrics used to measure the question in a measurable manner
● Usually used when you wanna cover different levels of organization
- Like goal usually is at the organizational level
- Questions are cascaded down to department level
- Metrics at cascaded to day to day/working level

- Narrative Best Practices


● Always know your audience
- Think about if they prefer storytelling or if he prefers to go point by point
● Clear central message
- When you craft the narrative, you need to have an idea what is the point you’re driving at
- Lecturer: Don’t get lost in the nitty gritty/technical details. Business people will ask so what?
● Setting
- Good well established setting creates intended mood + provide the backdrop & environment for your narrative
● Clear Structure
- Structure should be clear and logical
● Narrow Scope
- Narrow the scope. Don’t include every minor detail. Try to focus only on the important points so that the central
message doesn’t get diluted

- Presentation Style Considerations


● Resources

● Know your Audience


- Lecturer: Sometimes I do presentations, I will check the audience’s linkedin profile. If the audience has an
engineering background, I will use engineering analogies/terms. If audience is not there voluntarily, you may have to
cater for more activities to keep the audience engaged
● Considerations as a Speaker

- Data Storytelling & Verbal Communication


● Institutional Talk
- Audience in institution tend to orient to specific goals that are related to their institutional identities
- Their perception of their own ability to contribute is shaped by their institutional identity
● Figurative Language/Metaphors
- Usage of metaphors/idiomatic language to capture audience’s attention
- Helps audience understand something novel through relation w something that is well-understood
- Reflective of audience’s constructed reality as systems/goals/values we have reflect the metaphors we use daily +
unconsciously deploy

● Pacing
- Vary your speed to emphasize/provide clarity to different parts of your message
- Lecturer: Also consider how long your audience needs to digest your info when determining pacing

- Data Storytelling & Non-verbal Communication


● Body Posture - Best Practices

● Facial Expressions
- Lecturer: Show the appropriate facial expressions for the topic you are presenting

● Hand Gesture
- Lecturer: Avoid using too much hand gesture and be mindful of the context you are using it in

- Data storytelling - Presentation Style


● Lecturer: Choose your presentation style based on what you intend to achieve
- Data Storytelling - Infographics

● Best Practices

You might also like