Διπλωματική εργασία--Πανεπιστήμιο Μακεδονίας, Θεσσαλονίκη, 2025.
(EL)
Made available in DSpace on 2025-02-06T10:59:46Z (GMT). No. of bitstreams: 2
license_rdf: 1025 bytes, checksum: 84a900c9dd4b2a10095a94649e1ce116 (MD5)
MastorasIoannisMsc2025.pdf: 1458765 bytes, checksum: 2ca6c537733b0d474ba5a82f76e57f3a (MD5)
Previous issue date: 2025-02-06
(EN)
Submitted by ΙΩΑΝΝΗΣ ΜΑΣΤΟΡΑΣ (
[email protected]) on 2025-02-06T10:38:00Z
No. of bitstreams: 2
license_rdf: 1025 bytes, checksum: 84a900c9dd4b2a10095a94649e1ce116 (MD5)
MastorasIoannisMsc2025.pdf: 1458765 bytes, checksum: 2ca6c537733b0d474ba5a82f76e57f3a (MD5)
(EN)
Approved for entry into archive by ΕΛΙΣΑΒΕΤ ΧΑΝΤΑΒΑΡΙΔΟΥ (
[email protected]) on 2025-02-06T10:59:46Z (GMT) No. of bitstreams: 2
license_rdf: 1025 bytes, checksum: 84a900c9dd4b2a10095a94649e1ce116 (MD5)
MastorasIoannisMsc2025.pdf: 1458765 bytes, checksum: 2ca6c537733b0d474ba5a82f76e57f3a (MD5)
(EN)
CAR application is an integrated tool created for this thesis, which aims to help users deal
with the conventional process of preprocessing data and then prepare them for
interpretation prior visualization in an interactive R Shiny interface. As data processing
becomes increasingly automated in a variety of domains (e.g., healthcare, social
sciences, business sectors etc.), CAR is designed to make advanced statistical methods
and machine learning models readily accessible for users who lack extensive
programming or knowledge on statistics by democratizing the practice of data analysis.
CAR provides the following key features: Automated data cleaning and imputation
methods, customizable data visualizations and comprehensive suite of statistical &
machine learning algorithms. Main topics include abstracting the data analysis
workflow, ingesting and cleaning data, analyzing them with multiple statistical tests,
extracting features from several machine learning models or even creating visualizations
using custom animations only after a few lines. CAR helps us build a bridge for different
professions to solve real-time big data quality management and automated analysis
problems, improving the efficiency of research work in practice. Through performance
testing with a real-world dataset, the thesis evaluates CAR's functionality as well as its
usability and versatility. Overall, this work has the potential to offer immediate and direct
benefits in a way that bridges complex data science methodologies all the way over to
end-user-friendly data exploration of datasets thus enabling better access & policy
making. The creation of the CAR application is a big advance in disseminating state-of
the-art data analysis methods to more people who may not have deep technical or
statistical expertise. By automatically preprocessing data and analyzing them to derive
valuable and meaningful insights, CAR democratizes the field of data science enabling
experts as well as novices with little technical knowledge in deriving key takeaways from
their datasets which can be critical for evidence-based decision making across several
domains.
(EN)
University of Macedonia
