Interactive product visualisation (IPV) addresses the shortcomings of product visualisation on e-commerce sites, such as the absence of tactile information and the inability to directly examine products. Previous research has not addressed the influence of IPV on customer satisfaction. Therefore, the objective of this study was to address this gap by investigating factors that may influence customer satisfaction. The Stimulus-Organism-Response (SOR) model was used as a lens to further investigate the effect of stimuli of an IPV environment on the affective states of users as well as satisfaction as a response to stimuli. The furniture e-commerce industry and IKEA’s IPV tool were selected as the basis for this research. To gather data, an online survey with 5-point Likert scales was created based on previous research to discover the relationships between these factors. A total of 115 questionnaire responses were validated and analysed using multiple regression analysis. The results of the analysis indicated that interactivity, perceived ease of use, and entertainment had a positive influence on customer satisfaction as mediated by pleasure and dominance. Practical implications are presented to guide businesses that seek to implement IPV effectively.

3D city models have been widely used in different applications such as urban planning, traffic control, disaster management etc. Effective visualisation of 3D city models in various scales is one of the pivotal techniques to implement these applications. In this thesis, a framework is proposed to visualise the 3D city models both online and offline using City Geography Makeup Language (CityGML) and Extensible 3D (X3D) to represent and present the models. Then, generalisation methods are studied and tailored to create 3D city scenes in multi-scale dynamically. Finally, the quality of generalised 3D city models is evaluated by measuring the visual similarity from the original models.   In the proposed visualisation framework, 3D city models are stored in CityGML format which supports both geometric and semantic information. These CityGML files are parsed to create 3D scenes and be visualised with existing 3D standard. Because the input and output in the framework are all standardised, it is possible to integrate city models from different sources and visualise them through the different viewers.   Considering the complexity of the city objects, generalisation methods are studied to simplify the city models and increase the visualisation efficiency. In this thesis, the aggregation and typification methods are improved to simplify the 3D city models.   Multiple representation data structures are required to store the generalisation information for dynamic visualisation. One of these is the CityTree, a novel structure to represent building group, which is tested for building aggregation. Meanwhile, Minimum Spanning Tree (MST) is employed to detect the linear building group structures in the city models and they are typified with different strategies. According to the experiments results, by using the CityTree, the generalised 3D city model creation time is reduced by more than 50%.   Different generalisation strategies lead to different outcomes. It is important to evaluate the quality of the generalised models. In this thesis a new evaluation method is proposed: visual features of the 3D city models are represented by Attributed Relation Graph (ARG) and their similarity distances are calculated with Nested Earth Mover’s Distance (NEMD) algorithm. The calculation results and user survey show that the ARG and NEMD methods can reflect the visual similarity between generalised city models and the original ones. / QC 20100923 / ViSuCity Project

This paper presents PythonVis, a novel Virtual Reality (VR) software visualization prototype for program comprehension. The motivation for PythonVis is to leverage the affordances of VR and the debugger tool to support software developers' comprehension of novel software. An experimental study with follow up interviews was conducted using 10 participants, comparing PythonVis to a desktop setup. The results indicate that PythonVis could be useful for getting a better overview over a whole code base. Limitation are addressed and further studies are suggested. / Den här avhandlingen presenterar PythonVis, en ny prototyp av ett Virtual Reality (VR) visualiseringsverktyg för programförståelse. Motivationen för PythonVis är att utnyttja de förmågor som VR erbjuder, samt felsökarverktyg för att underlätta mjukvaruutvecklares förståelse för okända program. En experimentell studie med följande intervjuer genomfördes med 10 deltagare där PythonVis jämfördes med en traditionell dator. Resultaten indikerar att PythonVis kan användas för att få en bättre överblick av en hel kodbas. Begränsningar i studien är adresserade och framtida forskning är förslagen.

Climate change is now more of a reality than ever. Carbon budget provides an upper limit on the amount of carbon that can be emitted, for the earth to be within an accepted temperature rise. People are aware that eating less meat, flying less and reducing product consumption would decrease their emission, but they are not sure on how much less. Existing studies show that people are quite willing to accept the concept of carbon budget but there is no tool that visualises this information at an individual level.  This study investigated the features that must be included in a carbon budget tool when visualising an individual’s information to create awareness. An exploratory design approach was used in this study where the design was iteratively developed from user feedback. Four preliminary designs were first developed and discussed in a focus group. With the feedback from the focus group, one design was finalised and redesigned as a carbon budget tool. This was tested in a one-on-one user testing where users had to complete tasks using the tool. The study results provided evidence that personalised information, effective text, labels, interactive features, and clear and simple layout are important features that must be included when designing a tool to visualise an individual’s carbon budget. The participants also rated an increase in awareness of carbon budget concept after using the tool. / Klimatförändringarna är nu mer verkliga än någonsin. En koldioxidbudget anger en övre gräns för mängden utsläpp av klimatgaser som kan släppas ut för att hålla ökningen av jordens medeltemperaturen inom en acceptabel gräns. Många människor är medvetna om att lägre köttkonsumtion, färre flygresor och en minskad konsumtion av saker skulle minska deras utsläpp, men de är inte säkra på hur mycket varje ändring påverkar utsläppen. Befintliga studier visar att människor accepterar begreppet koldioxidbudget i sig, men det finns idag inga verktyg för att visualisera en koldioxidbudget på individuell nivå.  Denna studie undersöker vilka funktioner som behöver ingå i ett verktyg för visualisering av en individuell koldioxidbudget, med syfte att skapa medvetenhet hos användaren. En utforskande designmetod användes där verktygets design utvecklades iterativt med feedback från användare. Först utvecklades fyra preliminära designprototyper som diskuterades i en fokusgrupp. Med feedback från fokusgruppen vidareutvecklades en av prototyperna till en slutgiltig version. Detta testades i en användartestning där användare utförde uppgifter med verktyget. Studieresultaten visade att personlig information, effektiv text, etiketter, interaktiva funktioner och tydlig och enkel layout är viktiga funktioner som måste inkluderas när man utformar ett verktyg för att visualisera en koldioxidbudget för individuell användning. Deltagarna bedömde också att de uppnått en ökad medvetenhet om konceptet koldioxidbudget efter att ha använt verktyget.

Detta arbete är ett första steg i att erhålla ett koncept som kan användas inom industrin som fokuserar på att visualisera temperatur samt möjliggöra betraktning av dolda komponenter. På uppdrag av NEAVA AB, utvecklas ett koncept för att övervaka temperaturen på en dold komponent i en teoretisk modellmaskin. I större maskiner med flera dolda komponenter kan temperaturen mätas med hjälp av en eller flera sensorer till vilket ett visualiseringssystem kopplat. Det kan vara svårt att fysiskt se vad problemet är vid haveri då komponenterna är gömda inuti höljet, vilket ofta behöver nedmonteras för att upptäcka och åtgärda problemet. En litteraturstudie görs för att undersöka hur temperaturen på dolda komponenter visualiseras idag. Utgående från en hypotetisk maskinhall finns en maskin bestående av ett ytterhölje, en axel och 2 lager med respektive lagersäte inne i maskinen, samt en temperatursensor som mäter temperaturen i respektive lager på maskinen. Är arbetets syfte att med ett ”proof of concept” visa att det är möjligt att visualisera yttre som inre strukturer och temperaturen hos ett maskinelement som sitter dolt monterat på en större struktur, exempelvis värmeutvecklingen i en elektrisk motor. Med mål att framställa en 3D-modell där detta är möjligt. Modellen ska användas som ett visuellt hjälpmedel för att kartlägga temperaturförändring och visa var i modellen ett lager är på väg att överhettas i syfte av att planera ett maskinstopp för att undvika haveri. Modellen framställdes utifrån en konceptutvecklingsmetod och 3D-CAD i programmet Blender som är NEAVA AB:s huvudsakliga modelleringsprogram. Lösningen för att betrakta inre struktur var att tilldela ytterhöljet ett transparent material, som tillika gjordes i Blender. Temperaturvisualiseringen gjordes i spelmotor-programmet Unity, för att lättare skapa interaktiva funktioner. De använda lagren i arbetet är enkel-rads cylindrisk rullager. Med en rotationshastighet på 2600 varv per minut, erhålls en uppskattad livslängd på 6479,03 timmar och drifttemperatur på 228,85 °C. Resulterande modell har en justerbar transparent sida där den inre struktur tydligt kan betraktas och temperatur på lager visualiseras med ändrade färger enligt en temperaturskala. Modellen framtagen i detta arbete ska kunna användas som grund för vidareutveckling av konceptet, där en sensor och visualiseringsprogram tillkopplas. / This study is the first step to gaining a concept that is ready to use in industry, that focusing on visualizing temperature and enable monitoring of hidden components. At the request of NEAVA AB, develops a concept for monitoring the temperature in a hidden component in a theoretical model of a machine. In larger machines with several hidden components, the temperature can be measured with the help of one or more sensor to which a visualization system is connected. It can be difficult to physically see what the problem is in the event of a breakdown as the components are hidden inside the housing, which often needs to be disassembled to detect and fix the problem. A literature study is done to investigate how the temperature of hidden components is visualized today. Starting from a hypothetical engine hall in which a machine composing of a casing, shaft, two bearings with brackets and a temperature sensor which measures the temperature of the bearings. Is the purpose of this study to prove with a proof of concept, that it is possible to visualize outer and inner structure as well as temperature on a component which is mounted hidden in a lager structure, for example heat generation in an electric engine. With the goal of producing a 3D-model where this is possible. The model is to be used as a visualizing tool for mapping changes in temperature and show where in the model a bearing is on the verge of overheating for the purpose of schedule a operating stop in order to avoid damage The model was produced using a concept development method and 3D CAD in the program Blender, which is NEAVA AB´s primary modelling program. The solution for inspecting the inner structure was to apply a transparent material to the casing, which was done in Blender. The temperature visualizing was done in Unity, a game-engine program, which provides an easier option for creating interactive functions. The bearings used in this study are single row cylindrical roller bearings. With a rotation speed of 2600 rotations per minute, an estimated life rating of 6479,03 hours and operating temperature of 228,85 °C are obtained. The resulting model has an adjustable transparent side in which the inner structure can be inspected, and the bearing temperature is visualized by shifting colours according to a temperature scale. The model produced in this study will be used as a base for future development of the concept, where a sensor and visualization program are connected.

Ce mémoire part du constat que la visualisation est le parent pauvre de l'approche de modélisation à base d'agents: peu citée, peu mise en avant, elle constitue néanmoins, pour beaucoup de modélisateurs, non seulement leur premier point d'entrée vers la construction de modèles, mais aussi une façon de plus en plus prégnante, de concevoir, vérifier, voire valider des modèles de systèmes complexes. Le domaine de la modélisation à base d'agents, longtemps peu structuré, est en train de progressivement s'organiser en termes de méthodologies de conception, de communication, de description. Le succès du protocole ODD est là pour en témoigner: face à des demandes de représentation de systèmes de plus en plus complexes, les modélisateurs ont besoin de mettre de l'ordre dans leurs façons de modéliser et ils y parviennent. Cependant, il est étonnant de constater qu'aucune place n'est réservée dans ODD, ni dans aucune autre méthodologie approchante, à la façon de visualiser le modèle décrit ou à concevoir. Pour beaucoup de théoriciens de la modélisation, cette étape n'existe tout simplement pas, ou, si elle existe, est considérée comme le lointain produit dérivé d'un modèle déjà conçu, vérifié et validé. Pourtant, l'étude des pratiques de la modélisation à base d'agents révèle tout le contraire: l'aller-retour entre l'écriture et la visualisation d'un modèle fait partie intégrante du quotidien de nombreux chercheurs, comme en témoigne le succès de la plate-forme NetLogo. Cette visualisation, partie d'une démarche intégrée, permet de façon intuitive de vérifier et raffiner aussi bien les comportements individuels des agents que les structures collectives ou émergentes attendues. Poussée à l'extrême, cette pratique se rencontre aussi dans les démarches de modélisation et/ou simulation participative, où la visualisation du modèle sert de médiation entre acteurs et de support aux tâches, collectives, de modélisation. Absente des propositions méthodologiques, la visualisation de modèles à base d'agents se retrouve donc essentiellement délimitée et structurée par des pratiques individuelles, parfois partagées sous la forme de préceptes , mais rarement généralisables au-delà de la conception d'un modèle. Il existe pourtant des façons de visualiser, des manières de chercher à faire ressortir une information spécifique, des méthodes à suivre pour étudier visuellement une abstraction… / Information visualization is the study of interactive visual representations of abstract data to reinforce human cognition. It is very closely associated with data mining issues which allow to explore, understand and analyze phenomena, systems or data masses whose complexity continues to grow today. However, most existing visualization techniques are not suited to the exploration and understanding of datasets that consist of a large number of individual data from heterogeneous sources that share many properties with what are commonly called "complex systems". The reason is often the use of monolithic and centralized approaches. This situation is reminiscent of the modeling of complex systems (social sciences, chemistry, ecology, and many other fields) before progress represented by the generalization of agent-based approaches twenty years ago. In this thesis, I defend the idea that the same approach can be applied with the same success to the field of information visualization. By starting from the now commonly accepted idea that the agent-based models offer appropriate representations the complexity of a real system, I propose to use an approach based on the definition of agent-based visualization models to facilitate visual representation of complex data and to provide innovative support which allows to explore, programmatically and visually, their underlying dynamics. Just like their software counterparts, agent-based visualization models are composed of autonomous graphical entities that can interact and organize themselves, learn from the data they process and as a result adapt their behavior and visual representations. By providing a user the ability to describe visualization tasks in this form, my goal is to allow them to benefit from the flexibility, modularity and adaptability inherent in agent-based approaches. These concepts have been implemented and experimented on the GAMA modeling and simulation platform in which I developed a 3D immersive environment offering the user different point of views and way to interact with agents. Their implementation is validated on models chosen for their properties, supports a linear progression in terms of complexity, allowing us to highlight the concepts of flexibility, modularity and adaptability. Finally, I demonstrate through the particular case of data visualization, how my approach allows, in real time, to represent, to clarify, or even discover their dynamics and how that progress in terms of visualization can contributing,in turn, to improve the modeling of complex systems.

A learning analytics dashboard (LAD) is an application that illustrates the activity and progress of a user in a self-regulated, online learning environment. This tool mines source data to provide meaningful information that supports decision making and positively impacts learning behaviour. Research on this topic explores how learning activities and pedagogical goals are impacted by integrating LADs into learning and/or teaching environments. Currently, the majority of the research is centred around predicting student academic performance and identifying students that are at risk of failing. The popularity of integrating technology into educational practices has led to the adoption of LADs into learning management systems (LMS) or massive open online courses (MOOCs). The objective of this paper is to develop a concept for a standalone prototype LAD, for an Introductory Statistics course (STA 1000), to be potentially integrated into the University of Cape Town's (UCT) LMS, Vula. The dashboard aims to create and incorporate meaningful visualisations, that have the potential to primarily assist students as well as educators. Visualised information in the LAD aims to positively impact students to enhance and drive effective learning, which could consequentially aid educators. Additionally, the dashboard will aim to provide actionable feedback, derived from predictive modelling and course analytics, that positively impacts learning behaviour and identifies factors that the student could most effectively use to leverage their chances of passing and improve academic performance. Predictive analytics aim to identify academic factors, that a student has control over, such as course assessments and engagement variables, at certain time points in the academic semester and provide a useful course of action at those time points. Other than variables measured throughout the course, the predictive modelling takes certain prior academic information into consideration.

<p>In this dissertation, we study three important problems in modern deep learning: adversarial robustness, visualization, and partially monotonic function modeling. In the first part, we study the trade-off between robustness and standard accuracy in deep neural network (DNN) classifiers. We introduce sensible adversarial learning and demonstrate the synergistic effect between pursuits of standard natural accuracy and robustness. Specifically, we define a sensible adversary which is useful for learning a robust model while keeping high natural accuracy. We theoretically establish that the Bayes classifier is the most robust multi-class classifier with the 0-1 loss under sensible adversarial learning. We propose a novel and efficient algorithm that trains a robust model using implicit loss truncation. Our  experiments demonstrate that our method is effective in promoting robustness against various attacks and keeping high natural accuracy. </p> <p>In the second part, we study nonlinear dimensional reduction with the manifold assumption, often called manifold learning. Despite the recent advances in manifold learning, current state-of-the-art techniques focus on preserving only local or global structure information of the data. Moreover, they are transductive; the dimensional reduction results cannot be generalized to unseen data. We propose iGLoMAP, a novel inductive manifold learning method for dimensional reduction and high-dimensional data visualization. iGLoMAP preserves both local and global structure information in the same algorithm by preserving geodesic distance between data points. We establish the consistency property of our geodesic distance estimators. iGLoMAP can provide the lower-dimensional embedding for an unseen, novel point without any additional optimization. We  successfully apply iGLoMAP to the simulated and real-data settings with competitive experiments against state-of-the-art methods.</p> <p>In the third part, we study partially monotonic DNNs. We model such a function by using the fundamental theorem for line integrals, where the gradient is parametrized by DNNs. For the validity of the model formulation, we develop a symmetric penalty for gradient modeling. Unlike existing methods, our method allows partially monotonic modeling for general DNN architectures and monotonic constraints on multiple variables. We empirically show the necessity of the symmetric penalty on a simulated dataset.</p>

On the web, data visualisation through charts and diagrams can help present data in a more readable way. This is often done through the usage of JavaScript libraries. We experimented with 5 JavaScript data visualisation libraries to determine their respective performances and how each one scaled with increased data size. Our results will hopefully provide help with the selection of said libraries. The results show a significant difference in response times between all libraries for mostdata sizes, with only a few exceptions. Different exponential growths were also identified for all libraries, and the performance often varied greatly depending on chart type. Response time is not the only variable in performance measurements. Future research could cover other aspects, like memory consumption and rendering requirements. There were also times when the libraries did not render at larger data sizes, despite showing no errors, and further investigation behind this should be done.

What is good map design and how should information best be visualised for a human reader? This is a general question relevant for all types of design and especially for digital maps and various Geographic Information Systems (GIS), due to the rapid development of our digital world. This general question is answered in this thesis by presenting a number of principles and tips for design of maps and specifically interactive digital visualisation systems, such as a GIS. Furthermore, this knowledge is applied to the application dpPower, by Digpro, which present the tools to help customers manage, visualise, design and perform calculations on their electrical networks. The visualisation and design of the network was analysed together with the usage of two common background maps, GSD-Fastighetskartan by Lantmäteriet and Primärkartan by the municipalities, whose default appearances are defined by Digpro. The aim was to answer whether there is a more suitable design of the background maps and network to better complement the usage of dpPower and if so, what is the better design? When designing interactive systems that will later have various end-users, a user-centred design is important. Therefore, the initial step was to collect user inputs and feedback on the current design via customer interviews. This gave a set of user criteria for good map design of dpPower specifically. A study of existing relevant literature and previous work was also performed where several general key principles for good design could be identified. Finally, a comparison between the dpPower design and other existing map products, such as e.g. Google Maps and Eniro, was made where key similarities and dissimilarities were identified and discussed. These user criteria and design principles could be combined, both to present an answer to the general question “What is good design?” and to present a suggestion of new map appearance in dpPower. Key considerations in the new design suggestions were e.g. to have a toned down background map with all features in the same hue family. However, for GSD-Fastighetskartan the important convention of land classes, blue = water, green = vegetation \&amp; yellow = open land, should be kept. Colour combinations and contrast is the most important design element and since a design cannot be optimally adapted for all types of colour vision deficiencies, the suggestion is to separate the designs to specifically target user groups of different colour vision abilities. Important map information such as e.g. detailed road data should be kept while unnecessary features such as contour lines and polygon borderlines should be hidden. Text positions should also be considered. The results were evaluated both via a survey, distributed to users of dpPower, GIT-students and users with no previous experience of GIT or dpPower, and a seminar with employees at Digpro. The conclusions drawn from the evaluation was that the presented design suggestions and principles are good, but adjustments should be made. E.g. a use of yellow for low voltage cables, as suggested for Red-Green impaired, is perhaps not the best solution. The results present a good foundation for design of dpPower but more adjustments should be made based on the evaluation and then another evaluation can be performed. It would give an even better result.