prikaz prve stranice dokumenta Primjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje
  Download
PDF 778.09 KB
master's thesis
Primjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje
Split: University of Split, University of Split, Faculty of science, 2021. urn:nbn:hr:166:482854
Džal, Daniela
University of Split
Faculty of Science
Department of Mathematics

Institutional repository: Repository of Faculty of Science

Cite this document

Džal, D. (2021). Primjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje (Master's thesis). Split: University of Split, Faculty of Science. Retrieved from https://urn.nsk.hr/urn:nbn:hr:166:482854

Džal, Daniela. "Primjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje." Master's thesis, University of Split, Faculty of Science, 2021. https://urn.nsk.hr/urn:nbn:hr:166:482854

Džal, Daniela. "Primjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje." Master's thesis, University of Split, Faculty of Science, 2021. https://urn.nsk.hr/urn:nbn:hr:166:482854

Džal, D. (2021). 'Primjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje', Master's thesis, University of Split, Faculty of Science, accessed 30 November 2024, https://urn.nsk.hr/urn:nbn:hr:166:482854

Džal D. Primjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje [Master's thesis]. Split: University of Split, Faculty of Science; 2021 [cited 2024 November 30] Available at: https://urn.nsk.hr/urn:nbn:hr:166:482854

D. Džal, "Primjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje", Master's thesis, University of Split, Faculty of Science, Split, 2021. Available at: https://urn.nsk.hr/urn:nbn:hr:166:482854

Metadata
TitlePrimjena strojnog učenja u ocjenjivanju kakvoće mora za kupanje
Title (english)Application of machine learning on evaluation of beach water quality
AuthorDaniela Džal
MentorIvo Ugrina (mentor)
MentorIvana Nižetić Kosović (pomoć pri izradi, neposredni voditelj)
Committee memberIvo Ugrina (predsjednik povjerenstva)
Committee memberDamir Vukičević (član povjerenstva)
Committee memberJurica Perić (član povjerenstva)
GranterUniversity of Split
Faculty of Science
(Department of Mathematics)
Split
Defense date and country2021-01-29, Croatia
Scientific / art field,
discipline and subdiscipline		NATURAL SCIENCES
Mathematics
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
Kakvoća mora za kupanje procjenjuje se na temelju količine bakterija escherichia coli i crijevni enterokok u uzorku. S obzirom da mikrobiološka analiza traje nekoliko dana, takva informacija nije upotrebljiva kod obavještavanja javnosti o trenutnoj kvaliteti vode. Stoga je cilj ovog rada primijeniti predikcijske modele na navedenim podacima. U okviru rada izgrađene su dvije feed forward neuronske mreže za klasifikaciju kakvoće mora, svaka za po jednu vrstu bakterije. Novi se uzorak klasificira kao nezadovoljavajuć ako barem jedna neuronska mreža predvidi prekomjernu količinu bakterija. Model je verificiran usporedbom sa stvarnim mjerenjima na skupu za testiranje. Dobiveni modeli klasificiraju onečišćenje prema bakteriji e. coli s informednessom 69.5%, a prema crijevnom enterokokiju s 83.2%. Opisanim spajanjem ovih modela dobije se model koji klasificira nove uzorke s informednessom 64%. S obzirom na specifičnost problema, tešku predvidljivost količine bakterija i nebalansiranost podataka, dobiveni su rezultati iznad očekivanja, a daljnje bi se poboljšanje dobilo češće uzorkovanim podacima.
Abstract (english)
Beach water quality assessment is based on the amount of e. coli and intestinal enterococci in a sample. Microbiological analysis lasts up to few days and gained information is not applicable to informing the public on present conditions. The goal of this thesis is to apply predictive models on available data. Two neural networks have been trained, each classifying beach water quality based on only one bacteria type. New sample is classified as non-satisfactory if a high level of bacteria is predicted by at least one neural network. Model is evaluated by comparing predictions with microbiological results on test data. Trained models classify beach water quality with informedness 69.5% for e. coli, and 83.2% for enterococci. Merged model has informedness 64%. Considering unpredictability of bacteria levels and data imbalance, acquired results are above expectations and further improvements could be achieved by training the model on more frequently sampled data.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:166:482854
Study programmeTitle: Mathematics; specializations in: major in pure mathematics, major in computing, major in teaching, major in statistics and computing
Course: major in computing
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra matematike (magistar/magistra matematike)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2021-02-02 17:36:40