prikaz prve stranice dokumenta Usporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina
Access restricted to students and staff of home institution
master's thesis
Usporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina
2024. urn:nbn:hr:124:204226
Orlović, Tomislav
University of Split
Faculty of economics Split

Cite this document

Orlović, T. (2024). Usporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina (Master's thesis). Split: University of Split, Faculty of economics Split. Retrieved from https://urn.nsk.hr/urn:nbn:hr:124:204226

Orlović, Tomislav. "Usporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina." Master's thesis, University of Split, Faculty of economics Split, 2024. https://urn.nsk.hr/urn:nbn:hr:124:204226

Orlović, Tomislav. "Usporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina." Master's thesis, University of Split, Faculty of economics Split, 2024. https://urn.nsk.hr/urn:nbn:hr:124:204226

Orlović, T. (2024). 'Usporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina', Master's thesis, University of Split, Faculty of economics Split, accessed 19 October 2024, https://urn.nsk.hr/urn:nbn:hr:124:204226

Orlović T. Usporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina [Master's thesis]. Split: University of Split, Faculty of economics Split; 2024 [cited 2024 October 19] Available at: https://urn.nsk.hr/urn:nbn:hr:124:204226

T. Orlović, "Usporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina", Master's thesis, University of Split, Faculty of economics Split, Split, 2024. Available at: https://urn.nsk.hr/urn:nbn:hr:124:204226

Metadata
TitleUsporedba neuronskih mreža i ARIMA modela za predviđanje prinosa Bitcoina
AuthorTomislav Orlović
MentorTea Šestanović (mentor)
Committee memberBranka Marasović (predsjednik povjerenstva)
MBZ: 274372
Committee memberŠime Jozipović (član povjerenstva)
MBZ: 383123
GranterUniversity of Split
Faculty of economics Split
Split
Defense date and country2024, Croatia
Scientific / art field,
discipline and subdiscipline		SOCIAL SCIENCES
Economics
Quantitative Economics
Abstract
Primarni cilj ovog rada i istraživanja je usporediti neuronske mreže i ARIMA modele korištene za predviđanje prinosa Bitcoina, istražiti modele i metode korištene u predviđanju prinosa kriptovaluta, kao i čimbenike koji utječu na kretanje cijena. U empirijskom dijelu rada korišteni su ARIMA modeli koji se inače koriste za prognoziranje vremenskih nizova, a dobiveni rezultati uspoređeni su s rezultatima metoda prognoziranja temeljenih na jednostavnim matematičkim modelima mozga – umjetnim neuronskim mrežama. Neuronsku mrežu koja je prikladna za korištenje kod vremenskih nizova zovemo neuronska mreža autoregresije ili NNAR model. Korišteni interni faktori i makrofinancijski podaci od siječnja 2016. do lipnja 2024. podijeljeni su na podatke za treniranje neuronske mreže i podatke za testiranje prediktivnih sposobnosti neuronske mreže i ARIMA modela. Dok se prihvaća hipoteza da modeli s uključenim dodatnim varijablama imaju bolje prediktivne sposobnosti u odnosu na iste modele bez uključenih dodatnih varijabli, hipoteza da NNAR modeli imaju bolje prediktivne sposobnosti u odnosu na ARIMA modele se samo parcijalno može prihvatiti.
Abstract (english)
The primary goal of this paper is to compare neural networks and ARIMA models used to predict Bitcoin returns, explore the models and methods used in cryptocurrency return prediction, as well as the factors that affect price movements. In the empirical part of the work, ARIMA models, that are normally used for forecasting time series, were used, and the obtained results were compared to the results of forecasting methods based on simple mathematical models of the brain – artificial neural networks. A neural network that is suitable for use with time series is called an autoregressive neural network or NNAR model. The internal factors and macro-financial data from January 2016 to June 2024 were used and divided into data for training the neural network and data for testing the predictive capabilities of the neural network and ARIMA model. While the hypothesis that models with selected variables have better predictive capabilities compared to the same models without additional variables included is accepted, the hypothesis that NNARX has better predictive capabilities compared to ARIMA models is only partialy confirmed.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:124:204226
Study programmeTitle: Business Studies
Study programme type: university
Study level: graduate
Academic / professional title: sveučilišni magistar ekonomije (sveučilišni magistar ekonomije)
Type of resourceText
File originBorn digital
Access conditionsAccess restricted to students and staff of home institution
Terms of useLicence In copyright icon
Created on2024-10-01 17:44:19