Schiano Di Cola, Vincenzo (2022) Data Science methodologies for predictive analytics in Smart Cities. [Tesi di dottorato]

[img]
Preview
Text
SchianoDiCola_Vincenzo_34.pdf

Download (39MB) | Preview
[error in script] [error in script]
Item Type: Tesi di dottorato
Resource language: English
Title: Data Science methodologies for predictive analytics in Smart Cities
Creators:
CreatorsEmail
Schiano Di Cola, Vincenzovincenzo.schianodicola@unina.it
Date: 14 March 2022
Number of Pages: 196
Institution: Università degli Studi di Napoli Federico II
Department: Ingegneria Elettrica e delle Tecnologie dell'Informazione
Dottorato: Information technology and electrical engineering
Ciclo di dottorato: 34
Coordinatore del Corso di dottorato:
nomeemail
Riccio, Danieledaniele.riccio@unina.it
Tutor:
nomeemail
Mazzocca, NicolaUNSPECIFIED
Piccialli, FrancescoUNSPECIFIED
Date: 14 March 2022
Number of Pages: 196
Keywords: Data Analytics, Machine Learning, Knowledge Extraction
Settori scientifico-disciplinari del MIUR: Area 01 - Scienze matematiche e informatiche > INF/01 - Informatica
Area 09 - Ingegneria industriale e dell'informazione > ING-INF/05 - Sistemi di elaborazione delle informazioni
Date Deposited: 22 May 2022 21:39
Last Modified: 28 Feb 2024 11:01
URI: http://www.fedoa.unina.it/id/eprint/14412

Collection description

The goal of this PhD dissertation is to conduct academic and industrial research on Data Science in a variety of fields. An interdisciplinary approach was required to address today's scientific and societal challenges. A three-year training path applied Data Science to two Smart City application domains: Cultural Heritage (CH) and E-Health, with a focus on machine learning (ML) and knowledge graphs (KG). The first application is on classifying and forecasting visitor flow within a museum. By applying Machine Learning to the CH sector, the study examined a mixed dataset of numerical and categorical values. A framework for data processing and information extraction for clustering visitor behaviors was developed to save time. The dissertation then focuses on two e-health topics: healthcare booking prescriptions and image processing for biosensors. Prescriptions issued by general practitioners were modeled as a KG to help optimize government and local e-health services. This dissertation aimed to identify more patterns in data than a legacy dataset and thus make more accurate predictions. The final biosensor application recognizes point of interests in smartphone photos and uses machine learning algorithms to determine their chemical composition. The tool predicts the amount of a compound based on the liquid sample's luminescence. This dissertation's specific research questions concentrate around one question: how can Data Science help construct Smart Cities? This is addressed through a framework for analyzing people moving indoors, an extension of a legacy SQL database to a Knowledge Graph, and the building of a lab-on-hand proof of concept. All of this is accomplished through the use of a wide range of mathematical and software methods, such as machine learning (clustering and classification), image processing, and KG embedding. Python and R with Grakn, AmpliGraph, OpenCV, and scikit-learn have been utilized as toolkits. Among the most important contributions made by this thesis are: a data processing framework for clustering visitor behavior (CH domain); tools to help CH decision-makers better analyze visitor behavior and data clusters (both are critical aspects in any kind of ML and decision-making tools); a framework for KG data management and analysis; a framework for biosensors recognizes point-of-interests in smartphone images and uses machine learning algorithms to estimate a compound's concentration in a liquid sample.

Downloads

Downloads per month over past year

Actions (login required)

View Item View Item