Title: Visual Genesis: Evolving

Visualizations through Data-Driven

Design
Chapter 1

1. Introduction

1.1. Background

Data has grown exponentially, changing how information is viewed, analysed, and

communicated across fields (Liang et al., 2023). As data volume, velocity, and variety

increase, good visualisation approaches are essential (Kehrer & Hauser, 2013). Data

visualisation helps humans understand complex data by visualising it for exploration, pattern

discovery, and decision-making (Munzner, 2014).

Moreover, data visualisation originated in computer science, statistics, graphic design, and

cognitive psychology (Ware, 2019). Jacques Bertin (1983) and Edward Tufte (1983)

pioneered graphic data display. Bertin's major work, "Semiology of Graphics," presented

graphic semiology's methodology for building visual representations based on data attributes.

However, Tufte stressed the need for clarity and simplicity in data visualisation by

maximising the data-ink ratio and minimising chartjunk.

Data visualisation has evolved to address the difficulties of rising data source complexity and

diversity. Visual analytics, interactive visualisations, and dimensional reduction help analyse

large datasets (Tenenbaum et al., 2000). High-dimensional data is reduced to lower-

dimensional areas to reveal patterns and structures (Tenenbaum et al., 2000). Visual analytics

helps people make informed decisions with automated data analysis and interactive

visualisations (Keim et al., 2008). Users can filter, zoom, and brush data in interactive
visualisations (Yi et al., 2007). Interactive features enhance data exploration, hypothesis

formulation, and user experience (Endert et al., 2017).

Recently, data visualisation and machine learning have enabled innovation (Wu et al., 2024).

Machine learning automates visualisation design, improves visual analytics, and analyses

complex datasets (Wongsuphasawat et al., 2017; Edent, 2017). Recommender systems can

suggest visual representations using data features (Wongsuphasawat et al., 2017; Moritz,

2018). Deep learning models automatically name and categorise visual patterns, speeding up

complex dataset processing (Wu et al., 2024).

In addition, data visualisation user experience and engagement are becoming important due

to technology (Lam et al., 2012). Feedback and interaction data connect visualisations to

users' mental models and decision-making processes, improving usability and relevance.

Visualisation user preferences have been studied utilising eye-tracking, think-aloud, and

interaction logs. Knowledge makes visualising design easier to utilise. Evolutionary

algorithms also improve visualisation design (Siirtola & Mäkinen, 2005). Inspired by natural

evolution, these algorithms iteratively build and assess solutions, selecting the fittest for

reproduction and mutation (Eiben & Smith, 2003). Hence, Evolutionary algorithms can

optimise visual representations to maximise information transmitting, minimise visual clutter,

and follow design principles and user preferences.

1.2. Objectives

This project aims to create data-driven visualisations that adapt to user needs. This data

visualisation technique combines data analysis, and machine learning. Research goals

includes:
 1. Create a framework for generating and optimising visualisations using machine

learning and data-driven methods.

2. Integrate user feedback and interaction data into visualisation design to improve user

experience.

3. Assess the successful and usable communication of complicated information using the

developed visualisations.

1.3. Research Questions

 How might data-driven design strategies improve communication of complicated

information through visualisations?

 What machine learning algorithms and techniques can optimise visualisations based

on data properties and user needs?

 How can user feedback and interaction data improve the user experience in

visualisation design?

 What evaluation methods can evaluate the usefulness and usability of evolving

visualisations compared to old approaches?

1.4. Problem Statement

Data visualisation and machine learning have advanced, yet there are no systematic

approaches to generate data- and user-specific visualisations. Time-consuming manual

visualisation design iterations reduce data potential (Wongsuphasawat et al., 2015). Data

sources and complexity may outpace traditional visualisation design methodologies (Endert

et al., 2017).
Recently, machine learning has been used to automate and enhance data visualisation,

boosting user engagement and handling complex data (Gumelar, 2019). Lack of systematic

approach to gathering and integrating user feedback and interaction data to develop

visualisations (Schmidt, 2020). The disparity can prevent user-centered visualisations that

meet needs.

1.5. Research Gap

Unsystematic data visualisation, machine learning, and UX design generate data- and user-

specific visualisations. Repetitive manual visualisations devalue data (Wongsuphasawat et

al., 2015). Traditional visualisation methods may fail as data sources and complexity grow

(Endert et al., 2017). Despite separating data analysis and user experience, Lam et al. (2012)

argue machine learning aids visual analytics and automation. But, visibility may ignore user

input.

Manual complex design iterations are laborious. This approach can significantly lessen data

visualisation's influence on real-time decision-making situations that require immediate

insights (Eilemann, 2019). Traditional designs may suffer with data source variability and

complexity, needing automated solutions.

Visualisation design badly integrates user input and interaction data. Schmidt (2020)

indicates that many systems lack user inputs, making visualisations less intuitive. Another

study (Luo et al., 2020) also highlights that a major challenge with automatic visualization

systems is that they can provide visualizations without understanding the user's intent,

leading to users being misled. This provides an opportunity to work with an user certric data

visualizaltion. Hence, such gap allows interaction-driven user-centered design. Data study

underutilizes modern visualisation methods. Interactive visualisations can supplement basic

charting in exploratory data analysis tools (Wang et al., 2019). This gap implies that data

science tools need better visualisation to maximise their potential.

1.6. Significance and Contribution

This research will use machine learning to develop user-focused data-driven visualisations.

The proposed method will simplify visualisation design, remove manual iterations, and

improve complex information visualisation. Machine learning and data visualisation will

boost both fields and enable multidisciplinary collaboration. Integrated data analysis,

visualisation design, and user experience could assist scientific research, corporate

intelligence, and decision support system visualisation (Endert et al., 2017). Create

customised visualisations for user goals and data characteristics to clarify complex

information and improve decision-making.

Chapter 2

1. Literature Review

1.1. Data Visualization

Visualising data involves computer science, statistics, graphic design, and cognitive

psychology. Charts, graphs, and maps simplify complex data (Munzner, 2014). Data

visualisations use the human visual system's ability to collect and evaluate visual information

to facilitate data exploration, pattern recognition, and decision-making (Ware, 2019).

Researchers show data using various visualisation methods and criteria. The data-driven

visual representation method graphic semiology was invented by Bertin (1983). Tufte (1983)

advised raising the data-ink ratio and minimising chartjunk to simplify data presentation.

Recent data visualisation has altered due to data source complexity and diversity (Minch,

2023). Visual analytics, interactive visualisations, and dimensional reduction examine huge

datasets (Tenenbaum et al., 2000). Hasugian et al. (2023) examine dimension reduction

approaches as PCA, LDA, t-SNE, and UMAP for visualising high-dimensional and

complicated data and their ability to preserve important information.

1.2. Data-Driven Design

User research and data analysis drive data-driven design (Hartson & Pyla, 2012). It designs

using empirical data and user insights, not intuition or subjective choices. This method works

in UX, product, and marketing design. Visualisations communicate data and address user

goals (Wongsuphasawat et al., 2016). Analysing data structure, distribution, and linkages
helped designers choose the best graphics (Munzner, 2014). Design and user research may

improve visualisations. User interviews, usability testing, and think-alouds improve designer

visualisations (Lam et al., 2012).

1.3. Machine Learning in Visualization

Data visualisation design and analytics are automated by machine learning (Wongsuphasawat

et al., 2017). Deep learning, grouping, and dimensionality reduction shape visualisation. Data

features display recommender systems (Wongsuphasawat, 2017; Moritz, 2019). Pattern and

relationship visualisations are suggested by machine learning.

Machine learning aids pattern, anomaly, and decision assistance in visual analytics (Endert et

al., 2017). Labelling visual patterns speeds up complex dataset exploration and analysis in

deep learning models (Wu et al., 2024).

1.4. User Experience and Interaction in Visualization

User experience determines data visualisation success. Feedback and interaction data connect

visualisations to users' mental models and decision-making processes, improving usability

and relevance (Lam et al., 2012).

Through eye-tracking, think-aloud, and interaction logs, studies visualise user behaviour and

preferences. These tips simplify visualisation design. Interactive visualisations enable real-

time data modification (Yi et al., 2007). Filtering, zooming, and brushing modify

visualisations (Endert et al., 2017).

 1.5. Evolution Algorithms in Design

Siirtola and Mäkinen (2005) apply nature-inspired algorithms to visualisation design. These

algorithms iteratively create and evaluate solutions to pick the fittest for reproduction and

mutation (Eiben & Smith, 2015).

Siirtola and Mäkinen (2005) say evolutionary algorithms can optimise visual representations

for many design goals and constraints. Visualisation information can be improved without

visual clutter, design limits, or user preferences via an evolutionary algorithm. They

iteratively develop and analyse visualisations to find optimal or near-optimal solutions in a

large solution space.

Evolutionary methods can automate and optimise complex design tasks (Parmee, 2012). Data

visualisations can be customised using evolutionary algorithms, machine learning, and input

(Siirtola & Mäkinen, 2005).

Chapter 3

2. Research Methodology

2.1. Data-Driven Visualization Design Framework

A machine learning and user experience-based data-driven visualisation design approach is

proposed. The framework includes:

2.1.1. Data Preprocessing and Feature Extraction

Before processing input data, the framework will check for missing values, outliers, and

manipulations. The researcher will use imputation, outlier detection, and normalisation to

assure data quality and consistency.

The data will next be analysed for statistical attributes (e.g., mean, variance, skewness), data

kinds (categorical, numerical), and domain-specific characteristics. These traits will reveal

data structure and linkages to inform visualisation design (Munzner, 2014).

2.1.2. Visualization Generation and Optimization

The retrieved features will be used to build early visualisations using D3.js (Bostock et al.,

2011) and Vega-Lite (Satyanarayan et al., 2017). These visualisations will start optimisation.

According to objective functions and constraints, an evolutionary algorithm will evaluate and

evolve visualisations. Considerations for objective functions include:

• Visualisation accurately depicts data qualities, such as types, distributions, and

relationships (Munzner, 2014).

 • Assessment of information conveyed by visualisation, considering pre-

attentive processing, Gestalt principles, and cognitive load (Ware, 2019).

 The user experience refers to the usability, interactivity, and overall experience of

a visualisation, informed by user input and interaction data (Lam et al., 2012).

 User research preferences, visualisation best practices (Tufte, 1983; Bertin, 1983),

and domain-specific requirements may constitute constraints.

To find novel solutions, the evolutionary algorithm will use mutation and crossover (Eiben &

Smith, 2015). The fittest candidates will be picked for the following generation based on

objective functions and limitations.

2.1.3. User Feedback and Interaction

The evolutionary technique optimises visualisations using user feedback and interaction data.

Eye-tracking, think-aloud protocols, and interaction logs will record user preferences. Eye

tracking data can indicate user confusion or visual interest (Goldberg & Helfman, 2010).

Users can provide qualitative feedback and communicate their ideas with think-aloud

visualisations. Users filter, zoom, and brush visualisations in interactive logs can also be

helpful.

User data will change the evolutionary algorithm's objective functions and constraints to meet

user preferences. If users have trouble with visual encoding or interactivity, the system may

prioritise alternatives.

2.1.4. Evaluation and Validation

The evolving visualisations will be analysed and validated using various methods:
  Quantitative evaluation: Use controlled trials and user studies to assess visualisation's

data delivery and analytical capabilities. Task completion, accuracy, and subjective

ratings can evaluate visualisations (Plaisant, 2004).

 Interviewing, focus grouping, and surveying experts and end-users for qualitative

input. This will reveal visualisation strengths, weaknesses, and enhancement

opportunities (Lam et al., 2012).

 Compare evolved visualisations to domain expert and commercial tool methodologies

and traditional design models. This will show if the proposed technique outperforms

previous ones.

The data-driven visualisation design approach will improve with iteration and validation.

2.2. Iterative and Incremental Development

Iterative and gradual research will strengthen the framework. Steps in iterations include:

 Develop a framework component like data pretreatment, visualisation, or user

feedback.

 Test the component with controlled trials, user research, and experts. Testing the

component using synthetic or real-world datasets and collecting quantitative and

qualitative performance and usability statistics may be necessary (Han et al., 2022).

 Find optimisation opportunities such algorithm inefficiencies, human interface

restrictions, and data or domain gaps using results analysis.

 Improve the component using prior data. Changes may include algorithms, objective

functions, restrictions, features, and approaches.

  Assess system performance and effectiveness after integrating the improvement into

the framework.

This iterative process will improve the framework utilising the newest research and ideas. It

will also allow progressive component creation and testing, reducing system development

complexity and risk in one phase.

2.3. Collaboration and Interdisciplinary Approach

The research is interdisciplinary, thus collaboration with other experts is crucial. Consultating

data visualisation experts and practitioners to guarantee best practises and domain-specific

requirements. Scientific visualisation, information visualisation, and visual analytics

researchers, designers, and domain professionals may be consulted (Munzner, 2014; Ware,

2019).

 Working with machine learning experts to use cutting-edge techniques and methods.

Evolutionary algorithms, deep learning, and recommender systems experts may be

needed (Eiben & Smith, 2015; LeCun et al., 2015; Ricci, 2011).

 Used perception, cognition, and user-centered design with user experience and

cognitive psychologists. This will guarantee that data visualisations are intuitive and

engaging for end-users (Ware, 2019; Lam et al., 2012).

 Testing updated visualisations in real-world settings with domain experts and end-

users. Data visualisation is crucial to decision-making and communication in

healthcare, finance, and environmental science, therefore this may include working

with researchers, practitioners, or stakeholders (Munzner, 2014).

Interdisciplinarity brings multiple viewpoints, experience, and domain knowledge to

research, improving its efficacy.

 2.4. Ethical Considerations

Ethics will guide the research. This involves: • Ensure user data privacy and confidentiality in

accordance with data protection rules and institutional norms during research. Anonymous or

aggregated user data, safe data storage and transmission, and informed agreement from

participants may be needed.

 Follow ethical criteria for human subject research, as established by IRBs or research

ethics committees. Appropriate approvals, participant safety, autonomy, beneficence,

and justice must be observed (Resnik, 2015).

 Addressing potential biases and limits of created methodologies candidly. Bias audits,

limitations documentation, and explicit instruction for using and interpreting

developed visualisations may be needed (Mehrabi et al., 2021).

 Encouraging ethical and appropriate use of research findings, especially in sensitive

or high-stakes areas. Develop standards or best practices for deploying and applying

the data-driven visualisation design framework, as well as end-user and stakeholder

training and education (Dignum, 2018).

 Monitoring and analysing ethical implications throughout study, modifying method to

accommodate evolving problems. An ethics advisory board or regular consultation

with ethics and responsible innovation specialists may help (Schiff et al., 2020).

This study can assist build trustworthy and responsible data visualisation systems that

prioritise user privacy, fairness, and transparency by considering ethical considerations

throughout the research process.

 2.5. Evaluation Metrics and Benchmarks

The data-driven visualisation design method will be measured and benchmarked.

 Improved data visualisations lead to faster task completion, accuracy, and subjective

evaluations (Plaisant, 2004; Hart & Staveland, 1988; Brooke, 1996)

 Use DVS or PBEVisTech to assess visualisation quality and effectiveness (Amar et

al., 2005).

 Evolutionary visualisation surveys measure user satisfaction, engagement, and utility

(Lam et al., 2012).

 The proposed framework will be compared to professional and commercial

visualisation methods. Benchmarks include datasets and case studies.

 Involved experts and stakeholders in establishing domain-specific standards and

evaluation criteria for practical visualisations (Munzner, 2014).

Partner and stakeholder input will shape evaluation metrics. A complete framework review

and precise performance and impact measurements will follow.

Reference list

Aggarwal, C. (2017). Outlier Analysis. [online] Available at:

https://sadbhavnapublications.org/research-enrichment-material/2-Statistical-Books/

Outlier-Analysis.pdf.

Amar, R., Eagan, J., & Stasko, J. (2005). Low-level components of analytic activity in

information visualization. In IEEE Symposium on Information Visualization, 2005.

INFOVIS 2005. (pp. 111-117). IEEE.

Berlin, J. (1983). Semiology of Graphics–Diagrams Networks Maps. Trails. William J. Berg.

Madison: U of Wisconsin P.

Bostock, M., Ogievetsky, V., & Heer, J. (2011). D³ data-driven documents. IEEE

transactions on visualization and computer graphics, 17(12), 2301-2309.

Brooke, J. (1996). SUS: A ‘Quick and Dirty’ Usability Scale. Usability Evaluation In

Industry, pp.207–212. doi:https://doi.org/10.1201/9781498710411-35.

Dempster, A.P., Laird, N.M. and Rubin, D.B. (1977). Maximum Likelihood from Incomplete

Data Via the EM Algorithm. Journal of the Royal Statistical Society: Series B

(Methodological), 39(1), pp.1–22. doi:https://doi.org/10.1111/j.2517-

6161.1977.tb01600.x.

Dignum, V. (2018). Ethics in artificial intelligence: introduction to the special issue. Ethics

and Information Technology, [online] 20(1), pp.1–3.

doi:https://doi.org/10.1007/s10676-018-9450-z.
Eiben, A.E. and Smith, J.E. (2015). Introduction to Evolutionary Computing. [online]

Natural Computing Series. Berlin, Heidelberg: Springer Berlin Heidelberg.

doi:https://doi.org/10.1007/978-3-662-44874-8.

Eilemann, S. (2019). Parallel Rendering and Large Data Visualization. ArXiv.

doi:https://doi.org/10.5167/uzh-168812.

Endert, A., Ribarsky, W., Turkay, C., Wong, B.L.W., Nabney, I., Blanco, I.D. and Rossi, F.

(2017). The State of the Art in Integrating Machine Learning into Visual Analytics.

Computer Graphics Forum, 36(8), pp.458–486.

doi:https://doi.org/10.1111/cgf.13092.

Goldberg, J.H. and Helfman, J.I. (2010). Scanpath clustering and aggregation.

doi:https://doi.org/10.1145/1743666.1743721.

Gumelar, A. B. (2019, September). An Anatomy of Machine Learning Data Visualization.

In 2019 International Seminar on Application for Technology of Information and

Communication (iSemantic) (pp. 1-6). IEEE.

Han, J., Pei, J., & Tong, H. (2022). Data mining: concepts and techniques. Morgan

kaufmann.

Hart, S.G. and Staveland, L.E. (1988). Development of NASA-TLX (Task Load Index):

Results of Empirical and Theoretical Research. [online] ScienceDirect. Available at:

https://www.sciencedirect.com/science/article/pii/S0166411508623869.

Hartson, R. and Pyla, P.S. (2012). The UX Book: Process and Guidelines for Ensuring a

Quality User Experience. [online] Google Books. Elsevier. Available at:

 https://books.google.com/books?

hl=en&lr=&id=w4I3Y64SWLoC&oi=fnd&pg=PP1&dq=Hartson.

Hasugian, P.M., Mawengkang, H., Sihombing, P. and Efendi, S. (2023). Review of High-

Dimensional and Complex Data Visualization. 2023 International Conference of

Computer Science and Information Technology (ICOSNIKOM).

doi:https://doi.org/10.1109/icosnikom60230.2023.10364377.

Kehrer, J., & Hauser, H. (2013). Visualization and visual analysis of multifaceted scientific

data: A survey. IEEE transactions on visualization and computer graphics, 19(3),

495-513.

Keim, D.A., Mansmann, F., Schneidewind, J., Thomas, J. and Ziegler, H. (2008). Visual

Analytics: Scope and Challenges. Lecture Notes in Computer Science, pp.76–90.

doi:https://doi.org/10.1007/978-3-540-71080-6_6.

Lam, H., Bertini, E., Isenberg, P., Plaisant, C., & Carpendale, S. (2012). Empirical studies in

information visualization: Seven scenarios. IEEE transactions on visualization and

computer graphics, 18(9), 1520-1536.

LeCun, Y., Bengio, Y. and Hinton, G. (2015). Deep Learning. Nature, 521(7553), pp.436–

444. doi:https://doi.org/10.1038/nature14539.

Liang, R., Huang, C., Zhang, C., Li, B., Saydam, S., & Canbulat, I. (2023). The fusion of data

visualisation and data analytics in the process of mining digitalisation. IEEE Access.

Luo, Y., Qin, X., Chai, C., Tang, N., Li, G., & Li, W. (2020). Steerable self-driving data

visualization. IEEE Transactions on Knowledge and Data Engineering, 34(1), 475-

490.
Mehrabi, N., Morstatter, F., Saxena, N., Lerman, K. and Galstyan, A. (2021). A Survey on

Bias and Fairness in Machine Learning. ACM Computing Surveys, [online] 54(6),

pp.1–35. doi:https://doi.org/10.1145/3457607.

Minch, B. (2023). In search of the most efficient and memory-saving visualization of high

dimensional data. arXiv (Cornell University).

doi:https://doi.org/10.48550/arxiv.2303.05455.

Moritz, D., Wang, C., Nelson, G.L., Lin, H., Smith, A.M., Howe, B. and Heer, J. (2019).

Formalizing Visualization Design Knowledge as Constraints: Actionable and

Extensible Models in Draco. IEEE Transactions on Visualization and Computer

Graphics, 25(1), pp.438–448. doi:https://doi.org/10.1109/tvcg.2018.2865240.

Munzner, T. (2014). Visualization analysis and design. CRC press.

Parmee, I.C. (2012). Evolutionary and Adaptive Computing in Engineering Design. Springer

Science & Business Media.

Schmidt, J. (2020). Usage of Visualization Techniques in Data Science Workflows.

Proceedings of the 15th International Joint Conference on Computer Vision, Imaging

and Computer Graphics Theory and Applications.

doi:https://doi.org/10.5220/0009181903090316.

Siirtola, H. and Mäkinen, E. (2005). Constructing and Reconstructing the Reorderable

Matrix. Information Visualization, 4(1), pp.32–48.

doi:https://doi.org/10.1057/palgrave.ivs.9500086.
Tenenbaum, J.B. (2000). A Global Geometric Framework for Nonlinear Dimensionality

Reduction. Science, 290(5500), pp.2319–2323.

doi:https://doi.org/10.1126/science.290.5500.2319.

Tufte, E. R. (1983). The visual display of quantitative information.

Wang, J., Hazarika, S., Li, C. and Shen, H.-W. (2019). Visualization and Visual Analysis of

Ensemble Data: A Survey. IEEE Transactions on Visualization and Computer

Graphics, 25(9), pp.2853–2872. doi:https://doi.org/10.1109/tvcg.2018.2853721.

Ware, C. (2019). Information visualization: perception for design. Morgan Kaufmann.

Wongsuphasawat, K., Moritz, D., Anand, A., Mackinlay, J., Howe, B., & Heer, J. (2015).

Voyager: Exploratory analysis via faceted browsing of visualization

recommendations. IEEE transactions on visualization and computer graphics, 22(1),

649-658.

Wongsuphasawat, K., Qu, Z., Moritz, D., Chang, R., Ouk, F., Anand, A., Mackinlay, J.,

Howe, B. and Heer, J. (2017). Voyager 2: Augmenting visual analysis with partial

view specifications. Proceedings of the 2017 CHI Conference on Human Factors in

Computing Systems. doi:https://doi.org/10.1145/3025453.3025768.

Wu, Y., Wan, Y., Zhang, H., Sui, Y., Wei, W., Zhao, W., Xu, G. and Jin, H. (2024).

Automated Data Visualization from Natural Language via Large Language Models:

An Exploratory Study. [online] arXiv.org. doi:https://doi.org/10.1145/3654992.

Xu, R. and WunschII, D. (2006). Survey of Clustering Algorithms. IEEE Transactions on

Neural Networks, 16(3), pp.645–678. doi:https://doi.org/10.1109/tnn.2005.845141.

Yi, J. S., ah Kang, Y., Stasko, J., & Jacko, J. A. (2007). Toward a deeper understanding of the

role of interaction in information visualization. IEEE transactions on visualization

and computer graphics, 13(6), 1224-1231.