ITSS 4355: Data Visualization - Week 8 PDF

ITSS 4355: Data Visualization Class 8: Data Storytelling Change in Syllabus 2 Objectives 1. Explain the importance of knowing your audience’s needs and analytical comfort level to create an effective data visualization or presentation 2. Explain how to create empathy...

ITSS 4355: Data Visualization Class 8: Data Storytelling Change in Syllabus 2 Objectives 1. Explain the importance of knowing your audience’s needs and analytical comfort level to create an effective data visualization or presentation 2. Explain how to create empathy in the audience with the data to create the most effective message possible in your data visualization or presentation 3. List the types of data visualizations that are most appropriate to communicate specific insights and for audiences with different needs and different levels of analytical comfort 4. Use dot matrix charts and big associated numbers to give the audience a better relative understanding of large numerical values in a data visualization 5. Describe examples of how data visualizations can be created to help the audience empathize with data 6. Identify situations for which a slope chart is appropri­ate and be able to create a slope chart from data 7. Use pre-attentive attributes to emphasize certain insights for an audience and tell a story in a data visualization 8. Define Aristotle’s Rhetorical Triangle and explain how it can be employed to connect with the audience to tell a story in a data visualization or presentation 9. Define Freytag’s Pyramid and explain how this provides a suggested structure for telling a story in your presentation 10. Define the concept of storyboarding and explain how to storyboard a presentation using sticky notes and using PowerPoint 11. Create an effective story for a presentation that considers the needs of the audience Data Visualization Makeover Clustered Column Chart Slope Chart The slope chart communicates more effectively than the clustered column chart how youth participation in primary elections has increased in four of these five states between 2008 and 2016. The title also better captures the main message of the visualization. Storytelling Storytelling refers to the ability to build a narrative from the data that is meaningful and memorable and is likely to influence the audience. The first goal in effective storytelling is to ensure we understand our audience by determining: 1. The insight needs of the audience from our data visualization: High-level understanding Detailed understanding 2. The level of analytical comfort in our audience: Low analytical comfort High analytical comfort Audience Member Needs The Chan Life Insurance Group investment in new process technology Chan Life Insurance wants to provide its employees with updates on the new process. At Chan Life Insurance there are two major audiences of employees with different needs: Sales agents’ needs: a decreased time required to receive a response for an insurance policy quote request  high-level understanding. Underwriters’ needs: more details on how and why the new technology system will reduce the time required to respond to requests for quotes  detailed understanding. Need for High-Level Understanding A line chart for Chan Life Insurance sales agents’ audience that displays the total response time for quotes Sales agents  high-level understanding September and November projections are indicated with a dashed line. Follow these steps to change a portion of a line chart into a dashed line: Step 1. Double click an individual data point in the chart to open the Format Data Point task pane Step 2. Select the Fill & Line icon and change the Dash type under Line Need for Detailed Understanding A clustered column chart for Chan Life Insurance underwriters’ audience that displays the task times for responses to quotes Underwriters  detailed understanding The chart shows the projected impact of the new information technology system on each of the four tasks required to complete a request for a quote from a sales agent. This chart may warrant further conversations on the impact of the new process technology on specific tasks. Audience Member Analytical Comfort Levels Patient satisfaction survey results at General Hospital General Hospital presents regular updates on patient satisfaction survey results from surgical patients to its staff. The General Hospital staff consists of two main audiences with either a low or a high analytical comfort level. Low Analytical Comfort Level A column chart that displays average patient satisfaction scores for surgical patients at General Hospital Low analytical comfort level The sorted column chart gives a simple overview of the main results of the patient satisfaction scores. It also makes relative comparisons among the department easier for the audience: Neurologic, General, and Vascular departments have low average patient satisfaction scores. Ophthalmic, Thoracic, and Orthopedic departments have high average patient satisfaction scores. High Analytical Comfort Level A box and whisker chart that displays details on patient satisfaction scores for surgical patients at General Hospital High analytical comfort level The box and whisker chart provides more insights into the results of the patient satisfaction scores: The Thoracic and Orthopedic departments have very similar average patient satisfaction scores, but very different distributions. The Ophthalmic surgical patient satisfaction scores have little variability while there is considerable variability in the patient satisfaction scores from Pediatric, Vascular, and General Surgery patients. Best Choice of Data Visualization The Best Data Visualization Depends on the Needs and Analytical comfort Level of the Intended Audience What Helps the Decision-Maker? The two different clustered bar charts assist a decision-maker choosing among three competing new IT systems for Chan Life Insurance Company The two different ways to build a clustered bar chart make it easy for the decision-maker to compare the various IT software vendors on multiple criteria. It can also lead to further discussion so that the decision-maker can ask for more details on each subjective evaluation. Empathizing With Data Empathy refers to the ability to understand and share the feelings of others. By empathizing with the data, we can create data visualizations that impact decisions effectively. The two common challenges for creating empathy with data are: Audiences can lose the ability to associate meaning when considering large numerical values. It can be challenging to consider individual cases when looking at aggregate statistics. Strategies for Large Numbers Strategies for helping the audience to interpret large numbers Convert a large number into something with which the audience may be more familiar. Show a Powerball Lottery payout of $357 million, as $171,000 per week for the next 40 years. To convey exponential growth, replace the use of large numbers with sentences like “The last two years saw the creation of 90% of all the data in existence.” Avoid the use of scientific notation for large numbers in data visualization. Use words such as “millions” and “billions” rather than 10 6 and 109. It is difficult to convey the magnitude of change in charts without proper context (see an example in the following two slides.) Empathizing With Data: a Simple Line Chart A simple line chart that shows the monthly unemployment rate in the United States Between 1948 and 2020 Near the end of the line chart, you can see the effect of the COVID-19 pandemic on unemployment in the United States. The unemployment rate has increased substantially but it can be difficult for the audience to interpret the magnitude of this change without additional context. Empathizing With Data: a Dot Matrix Chart A dot matrix chart that illustrates the number of jobs lost in the United States during the 2020 COVID-19 pandemic relative to the total labor force in the state of Ohio To add context to a visualization, consider using a dot matrix chart. A dot matrix chart is a simple chart that uses dots (or another simple graphic) to represent an item or groups of an item. The dots are laid out in a matrix form. The size of the matrix is relative to the size of the total number to be conveyed. *See step-by-step instructions in the notes on how to create a dot matrix chart in Excel. Empathizing With Data: Add a Picture A stacked column chart that includes a picture of a specific shelter dog creates more empathy with the data Consider creating empathy with the data in the audience by focusing on the individual rather than just presenting aggregated statistics. To add a picture to an existing chart in Excel, click the Insert tab on the Ribbon, click the Pictures button in the Illustrations group, and then choose from where we want to take the picture. The picture used in this chart comes from Online Pictures. Empathizing With Data: a Clustering Algorithm Summary results from a clustering algorithm used for the customers of Third State Bank In market segmentation analysis, clustering algorithms divide a company’s customers into different groups (clusters) that have similar characteristics. To create the best possible story, Third State Bank (TSB) identified cluster #7 of its customer segmentation analysis with Sophia; a fictional character persona meant to create empathy in the audience: “Sophia is a young single female who lives in a downtown apartment. Sophia is well-educated, follows the news closely, is likely to be politically active, and does not regularly attend sporting events.” Choosing the Correct Chart to Tell Your Story Data collected to Analyze Student A clustered column chart for student Enrollment at Opening Horizon Locations enrollment data at Opening Horizon Opening Horizon is interested in a year-over-year comparison of student enrollments for each location. However, the insight that Durango and Salida enrollment has moved in different directions compared to last year is not apparent to the audience. Build a Slope Chart A slope chart for the student enrollment data at Opening Horizon To create a slope chart in Excel, (A) build a simple line chart, (B) switch rows and columns, and (C) format the chart to its final version. *See step-by-step instructions in the notes. Using the Color Attribute to Tell Your Story The preattentive attribute of color emphasizes the impact the 2008 subprime mortgage crisis had on housing prices in Florida Many states are represented on the line chart to the left, with many possible insights. It is difficult for the audience to interpret any particular insight because the chart is complicated. The chart to the right uses the preattentive attribute of color to focus on the state of Florida. *See step-by-step instructions in the notes. Using the Size Attribute to Tell Your Story A chart the incorporates the use of a BAN to emphasize the number of dogs that enter animal shelters each year in the United States One simple but effective use of size as a preattentive attribute in data visualization is using a big associated number (BAN). We can illustrate the use of a BAN to emphasize the approximate number of dogs that enter animal shelters each year. A BAN can be added to an existing chart in Excel by clicking the Insert tab on the Ribbon, clicking the Text button, and selecting Text Box to add a text box to the chart. The BAN should havea large font size. Aristotle’s Rhetorical Triangle The Rhetorical Triangle describes the three ways in which a compelling story should connect with an audience A Rhetorical Triangle represents Aristotle’s three general ideas in which a story should connect with the audience. Ethos represents the ability to show credibility in a story to your audience. Logos typically refers to the logic or reasoning in the story or presentation. Pathos refers to the idea of connecting with the audience using emotion. Freytag’s Pyramid Freytag’s Pyramid outlines the common structure of effective stories A Freytag’s Pyramid divides a story into five distinct elements: 1. Introduction: it presents background information for the audience to connect to the story. 2. Rising action: begins to outline more details about the major conflict in the story. 3. Climax: the audience is exposed to the major conflict in the story. 4. Falling action: the audience starts to anticipate how the story will reach its conclusion. 5. Conclusion: also called denouement or resolution, it presents the end of the story. Application of Freytag’s Pyramid to Storytelling The Freytag’s Pyramid provides a better Hawaiian Bell presentation to Senior story to communicate the insights Team about customer service issues The Introduction includes the basics of the problem and why it is essential, by Presentation is mistakenly made in introducing a fictional protagonist, Liann, who the following chronological order: is facing an internet outage. The Rising action describes Liann’s struggle 1. A detailed explanation of the adopted data cleaning efforts. with customer service to have the internet outage fixed. 2. An alternative explanation of the The Climax develops the conflict between exploratory data analysis, Liann and the company as the internet including descriptive statistics. outage issue is diagnosed and resolved. 3. The aggregated findings from The Falling action uses the narrative to these data. provide recommendations on improving Issue: this approach is usually not customer service levels in the company. the most effective in terms of telling The Conclusion outlines the costs and an effective story. benefits of the recommendations and points out any important assumptions or any known limitations in the analysis. Storyboarding A storyboard is a simple visual organization of the main points of the story used to provide the structure of the narrative that you intend to create for the audience There are two standard methods for creating storyboards: (1) a low-tech method using sticky notes or (2) a higher-tech method using presentation software. A partial storyboard for the Hawaiian Bell A partial storyboard for the Hawaiian Bell presentation that uses sticky notes presentation that uses PowerPoint Check Your Knowledge 1. For good storytelling, what type of title should be used to involve the reader, make the insights memorable, and persuade the audience to take action? a. A descriptive title b. A predictive title c. An arbitrary title d. No title 2. A simple bar chart is effective in giving a straightforward overview of the results of patient satisfaction scores and is best suited for __________. a. an audience with low comfort with analytics b. an audience with high comfort with analytics c. any audience d. no audience Data Storytelling: Avoid the 10 Common Mistakes 1. Not knowing the difference between exploratory vs explanatory analysis 2. Not understanding your data 3. Choosing the wrong charts and visualizations 4. Choosing a confusing color scheme 5. Choosing an inappropriate report medium 6. Not providing enough context for your audience 7. Not knowing your audience well enough 8. Not providing commentary or informed analysis of performance 9. Not providing a way out of a catastrophe or a way to capitalize on success (missing next steps) 10. Being afraid of the conversation Some Resources 1. 11 Examples of Good and Bad Data Storytelling — Juice Analytics 2. Why storytelling is so powerful in the digital era | Ashley Fell | TEDxUniMelb (youtube.com) 3. Hans Rosling's 200 Countries, 200 Years, 4 Minutes - The Joy of Stats - BBC Four (youtube.com) 4. 7 Effective Tips for Presenting Data at Work! (youtube.com) 5. PowerPoint Storytelling: How McKinsey, Bain and BCG create compelling presentations ( youtube.com) 6. 10 Common Data Storytelling Mistakes and How to Avoid Them - LAZARINA STOY. Final Project (10% of Final grade) 1. Interactive Resume 2. YOUR (Data) Story 3. Some examples: Interactive Resume Gallery (tableau.com) Javier's Interactive Resume | Tableau Public Resume - Simone Magagnini | Tableau Public Maria Brock - Interactive Resume | Tableau Public 4. Start thinking about how to model the underlying data 5. Sketch Summary The characteristics of the audience may affect which types of data visualizations are most effective. To explain the importance of being able to empathize with data so that you may create the most effective visualizations and presentations. To use pre-attentive attributes such as color and size to highlight particular insights from data. To build several new types of data visualizations such as the dot matrix chart and the slope chart. To illustrate the importance of storytelling and provide suggestions for structuring a presentation using concepts such as the Rhetorical Triangle and Freytag’s Pyramid. To create storyboards using either sticky notes or a presentation software such as PowerPoint. Attendance Next Class….Quiz Data Dashboards