The Multidimensional Visual Analyser (MVA)

Golob, Ožbej (author), Smrdel, Aleš (mentor), Andrews, Keith (comentor)

The Multidimensional Visual Analyser (MVA) is an open-source web application for visually exploring and analyzing multidimensional datasets. MVA combines the best features from the reviewed multidimensional visual analysis software and implements them as both a web and desktop application. It provides four synchronized visualizations: scatterplot matrix, scatterplot, similarity map, and parallel coordinates, together with labeled partitions (classes) and a table view. In particular, the parallel coordinates implementation is highly interactive and also displays categorical dimensions. The application is built with Node, TypeScript, Svelte, Flowbite, D3, and Three.js. By rendering records withWebGL, MVA maintains its performance even when handling rather large datasets. MVA can be built and deployed as a web application, and can also be built as a desktop application through Tauri. This thesis first reviews popular multidimensional visual analysis approaches and existing software tools. It then surveys modern web technologies, before describing MVA in detail. Two appendices provide a User Guide and Developer Guide.

• Language: English
• Keywords: web application, analysis tool, visual analyser, information visualization, multidimensional datasets, scatterplot matrix, scatterplot, similarity map, parallel coordinates, labelled partitions, Node, JavaScript, TypeScript, Svelte, Flowbite, D3, Three.js
• Work type: Master's thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2024
• PID: 20.500.12556/RUL-162710
• Publication date in RUL: 26.09.2024

Secondary language

• Language: Slovenian
• Title: Multidimenzionalni vizualni analizator (MVA)

Abstract

Multidimenzionalni vizualni analizator (MVA) je odprtokodna spletna aplikacija za vizualno raziskovanje in analizo multidimenzionalnih podatkovnih zbirk. MVA združuje najboljše lastnosti pregledane programske opreme za multidimenzionalno vizualno analizo in jih implementira kot spletno in namizno aplikacijo. MVA ponuja štiri sinhronizirane vizualizacije: matriko razpršenih diagramov, razpršeni diagram, mapo podobnosti in paralelne koordinate, skupaj z označenimi particijami (razredi) in tabelo. Implementacija paralelnih koordinat je zelo interaktivna in prikazuje tudi kategorične dimenzije. Aplikacija je razvita z Node, TypeScript, Svelte, Flowbite, D3 in Three.js. Z upodabljanjem primerkov z WebGL aplikacija MVA ohranja zmogljivost tudi pri obravnavi precej velikih podatkovnih zbirk. MVA je mogoče namestiti kot spletno aplikacijo, prek programa Tauri pa je mogoče namestiti tudi namizno aplikacijo. V tem magistrskem delu so najprej pregledani priljubljeni pristopi multidimenzionalne vizualne analize in obstoječih programskih orodij. Sledi pregled sodobnih spletnih tehnologij, nato pa podroben opis MVA. Dva dodatka vsebujeta uporabniški priročnik in priročnik za razvijalce.

• Keywords: spletna aplikacija, orodje za analizo, vizualna analiza, vizualizacija informacij, multidimenzionalni podatki, matrika razpršenih diagramov, razpršeni diagram, mapa podobnosti, paralelne koordinate, označene particije, Node, JavaScript, TypeScript, Svelte, Flowbite, D3, Three.js