chart 2 design process
About Jack Zheng
Associate Professor at Kennesaw.edu
Data
Chart
http www
visual
Design
Chart type
Data visual
Design Process
Chart Design Process IT 7113 Data Visualization J.G. Zheng Fall 2023 http://idi.Kennesaw.edu/it7113/ Content Overview This lecture notes discuss some chart design issues. • Chart design process and considerations • Tool selection 2 Why do we need a design process? • A design process involves a defined set of design considerations and tasks. – A process consists of steps arranged in an order. – A repeatable and defined design process embodies maturity in design capability and experience. • Benefits: – serves as a guide and a checklist to plan and manage the whole project – reduces the randomness and improves efficiency 3 Example Processes • There are various ways to define a design process or a list of consideration. – Each process consists of configurable steps and actions. – A process can be flexible • For example: – A 3-Step Approach To Data Visualization https://digitalimpact.io/getting- started-a-3-step-approach-to-data-visualization/ – A 5-step guide to data visualization https://www.elsevier.com/connect/a-5- step-guide-to-data-visualization – The Data Visualization Design Process: A Step-by-Step Guide for Beginners https://depictdatastudio.com/data-visualization-design-process- step-by-step-guide-for-beginners/ – Andy Kirk’s 4 stages: https://www.youtube.com/watch?v=GVkXbQOzKNs&t=754s or Andy Kirk’s book “Data at Work” Chapter 2 – Design process for information visualization https://www.interaction- design.org/literature/article/how-to-design-an-information-visualization 4 A Basic Chart Design Process Requirement analysis • Set goals, objectives, messages Chart type choice • Choose one basic chart type (general or industry specific) based on a number of factors (mainly purposes and data features) Representation design • Determine visual data coding, involving visual mapping and visual properties (SCOPeS) Presentation design • Apply perceptual and attention shaping best practices (for example, pre-attentive processing and Gestalt principles) to make charts more effective and efficient 5 The following is a basic simple process for the most often scenario: We need to visualize all data using a commonly used chart type (or with limited alterations). Details of each step/task are presented in following slides. 1. Goals and Contexts • Before any design or technical work. Analyze the requirements and be clear about the following factors, which impact all following steps. • Key questions – What is the general goal and purpose of the data visualization? – What message am I trying to communicate with the data? – Who is the audience? 6 General Charting Purposes Purpose/function Description Comparison Comparing and sorting data points; can also compare to benchmarks or norms. Composition A hierarchy relationship. Also, it may imply part-to-whole comparisons. Distribution Aggregated value (usually count) of data points placed in categories; the category can be value ranges or time (trend). Relationship How things (data items) are related or positioned in a bigger context. Trend/evolution Variation of comparison involving temporal data. Profiling To comprehend things through visual shapes and patterns. 7 Review the six general purposes or categories of charts in module 3 – https://www.qlik.com/blog/third-pillar-of-mapping-data-to-visualizations-usage (the basic four) – http://www.excelcharts.com/blog/classification-chart-types/ (added evolution and profiling) Messages vs. General Purposes • What message or point I want to make? – Messages are more granular details, insights, or stories you want to emphasize. – Purposes focuses on big picture for exploration while messages focus on more specific insights for communication. – The latest story telling trend in data visualization focuses more on the message. • Example – General purpose: compare student enrollments in different degree programs across three departments – One may want to deliver different messages: • to emphasize the size of MSIT as the biggest program, or • to show in-significance among programs/departments • Another example: 8 https://www.storytellingwithdata.com/blog/2012/10/my- penchant-for-horizontal-bar-graphs A general purpose on comparison among areas A more specific message or point Audience • Consider specific user needs and preferences – User’s familiarity of certain charts or techniques • Some chart types may be more complicated for a less-experienced person to understand, but they can communicate information better for more advanced users. For example, radar chart or dual axis chart. – Personal preferences: some may prefer more condense visual format while other prefer guided story style with narratives – Corporate culture – Business sector/industry • These considerations will influence the choice of more specific chart type and presentation style. • For example, users (audiences) from a certain industry may have a convention and expectation of using certain color scheme, type of charts, and layout. 9 2. Chart Choice • Choosing the specific chart is also depends on the following important considerations. 1. Start with some general grand purposes; match the intended purpose and the chart’s major function. – Please review slide #7 and module 3 for more details. – Grand purposes and categories still may lead to several choices; so, this is just a starting point. 2. Examine the data to be presented. – What kind of data? Need to know data types, structures, and # of attributes and data items. • Other considerations (not focused on in this lecture) – Shape and size of charting space might be a factor – User/audience needs, preferences, conventions, etc. – Emotion, affection, culture, etc. – Chart customization: is one chart enough? 10 Initial Selection of Chart Type 11 • Match the intended purpose and the chart’s major function. – Many guides and tools are created to guide the selection of chart types based on purposes. Some of them are one-page quick references and others are more interactive and detailed. – Please review module 3 for more details. Abela's version • https://extremepresentation.com/de sign/7-charts/ • https://www.qlik.com/blog/third- pillar-of-mapping-data-to- visualizations-usage Camões’s version https://excelcharts.com/classification- chart-types/ Juice Analytics https://www.juiceanalytics.com/chartc hooser Schwabish’s Graphic Continuum • https://policyviz.com/2014/09/09/graphi c-continuum/ • https://www.informationisbeautifulawar ds.com/showcase/611-the-graphic- continuum Financial Times Visual Vocabulary https://www.ft.com/vocabulary Fraconeri’s version http://experception.net Ferdio* http://datavizproject.com Data catalog* http://www.datavizcatalogue.com From Data to Viz https://www.data-to-viz.com Chart make directory http://chartmaker.visualisingdata.c om Examine Data Set Features • Important data features to consider – Data size: how many data items or data serials will be in the chart? – Data type: temporal data, geo data, textual data, performance data, etc. • Data size: different chart types have different limit for data sizes. Understand each chart type’s limit. – How many data points or items (rows of data)? – How many dimensions or attributes (variables)? • Variable/measure type – including data value range, measuring unit, business meaning, etc. – Measure vs. dimension – Numerical, ordinal, and nominal (categorical) – Are they same in range and unit? – Are data needed to be presented of consistent type of mixed? • Data structure – Is there any specific relationship between data items and between variables, such as hierarchical? • In addition, data model – Is the source data in a format suitable for the intended visual/chart? – Do we need to structure/transform data set in a particular form for desired visualization? – Depending on the tool, there may be a specific way to structure and transform the data, or special calculations need to be performed. For example, waffle chart in Tableau https://www.pluralsight.com/guides/tableau-playbook-waffle-chart 12 Data Size (Variables) Impact on Chart Choice 13 See how data set features are described at https://www.data- to-viz.com and https://datavizproject.com (under the “Input” menu item) Number of variables leads to different chart types Data Size (Items) Impact on Chart Choice • A few data items <5 – Pie chart, column/bar chart • Quite some items <10 or 20 – Bar chart, line, bubble chart • Many many data items (rows) >20 – Profile charts, tree map, bubble chart, scatterplot, line charts (many time periods), parallel coordinates, etc. 14 Pie charts cannot handle too many data points; and can only display one variable. Bubble charts can handle up to 4 variables, and many data items. 3. Representation (Visual Features) Design • A chart type only provides a foundation or framework to the final chart design. The next step is to apply various visual features. • Note a chart type already sets a framework for most of the visual mapping on measures – column chart maps values to size of the columns, etc. This usually involves two major tasks. 1. Visual mapping – mapping of data to visual elements of the chosen chart; this is usually chart specific – For example, in a dual axis combo chart, which data serials should be mapped to x axis or y axis, etc.; which data should be represented by bars, etc. – Or in a cluster column chart, determining clustering order. – Or in a bubble chart, choose which dimension to be coded (as the bubble chart support limited number of dimensions) 2. Choice of visual property and visual encoding for data – SCOPeS (refer to module 2 https://www.edocr.com/v/631d1wpb/jgzheng/scopes-visual- properties). For example, choosing colors or sizing options. – Proper use of visual variables; apply the right visual decoration/properties. – Color choices 4. Presentational Design • The main focuses of presentational design is the usability and perceptual enhancement features, which involves preattentive features and Gestalt laws • Apply pre-attentive attributes to distinguish the major data points or serials, or ones closely related to you message. – Pre-attentive processing can help to rapidly draw the focus of attention to a target with a unique visual feature – For example, using contrast to differentiate the part that needs to draw attention – http://kenhirakawa.com/significance-of-contrast/ – https://www.coursera.org/lecture/dataviz-design/strategic-use-of- contrast-sDV6C • Apply Gestalt principles to group and sort data points and chart objects • https://www.webfx.com/blog/web-design/data-visualization-gestalt-laws/ 16 For details, refer to learning module 2 https://www.edocr.com/v/e6ql9njn/jgzheng/dat a-visual-foundation Other Presentational Issues • Contextual data – Benchmark, trend line, total, average, difference, quartile, estimate, confidence range, etc. – How to add them? • added directly in the chart, as a data serial • using annotations • Other UI and decorative features – font, grid line, shading, etc. • Descriptive and explanatory features – including title, legend, annotation, label, etc. 17 Note: these issues are not focused on in this class. We will briefly mention general UI design principles and guidelines in module 5. Please apply them based on your own study and experience. Need Customization? • Most of the time, we start with a conventional and standard chart, and that may be enough. • Sometimes, we have some need to go beyond conventional chart types. • Customization examples – Add more objects and properties – Incorporate features from other chart types – Using additional charts • Need to stack or overlay additional chart to make desired effect • These additional charts can be hierarchical, supplementary, or just change of view/perspective. • Arrangement of multiple charts: need to consider position, layout, sequence, transition. • Multiple charts can be placed side by side, stacked, overlapped, and sequenced. • If, none of the standard chart and customization is satisfying – rare situation but it happens – Design some unique and innovative visualizations beyond common well- known chart types – a good research topic. 18 Other Design Issues • Charting principles and best practices (see module 5) • Design with geo data and map (see module 7) • Dashboard design (see module 9) • Interactivity features (see module 10) 19 Choosing and Using a Visualization Tool • Modern visualizations are largely dependent or enabled by visualization software tools • The design of the visualization should also consider tool features and capabilities • Know tool (software application) features – Data handling features: complexity, volume, structure, source import, calculation, data modeling/structuring and transformation capability, etc. – The possibility or complexity of certain types of charts – Customization and hacks: overlapping, static add, scripting, annotation – Data update: how, frequency – Delivery medium: screen type, web, etc. – Skills sets and needs of users and developers • Refer to this web app for a summary of tools and features match – http://chartmaker.visualisingdata.com 20 A Design Case 21 https://youtu.be/GVkXbQOzKNs?t=2152 The discussion of the case started at 35:52 More Cases • What’s your point? Communicate the most important insight, from a story telling perspective – https://www.storytellingwithdata.com/blog/2021/1/10/lets-improve-this-graph-yt9xj • An alternative to tree maps – https://www.storytellingwithdata.com/blog/2018/6/5/an-alternative-to-treemaps Key Resources • A 3-Step Approach To Data Visualization https://digitalimpact.io/getting-started-a-3-step- approach-to-data-visualization/ • A 5-step guide to data visualization https://www.elsevier.com/connect/a-5-step- guide-to-data-visualization • The Data Visualization Design Process: A Step- by-Step Guide for Beginners https://depictdatastudio.com/data-visualization- design-process-step-by-step-guide-for- beginners/ 23 Additional Good Resources • Andy Kirk’s 4 stages: https://www.youtube.com/watch?v=GVkXbQOz KNs&t=754s or Andy Kirk’s book “Data at Work” Chapter 2 • Design process for information visualization https://www.interaction- design.org/literature/article/how-to-design-an- information-visualization • Accurate vs. Emotional Comparisons https://www.datarevelations.com/accurate-vs- emotional-comparisons-sometimes-pies- bubbles-and-waffles-are-the-better-choice/ 24