prikaz prve stranice dokumenta Prognoza vremenskih serija korištenjem radnog okvira Apache Spark
No public access
master's thesis
Prognoza vremenskih serija korištenjem radnog okvira Apache Spark
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. urn:nbn:hr:168:794093
Bošnjak, Dario
University of Zagreb
Faculty of Electrical Engineering and Computing

Institutional repository: FER Repository

Cite this document

Bošnjak, D. (2019). Prognoza vremenskih serija korištenjem radnog okvira Apache Spark (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:794093

Bošnjak, Dario. "Prognoza vremenskih serija korištenjem radnog okvira Apache Spark." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. https://urn.nsk.hr/urn:nbn:hr:168:794093

Bošnjak, Dario. "Prognoza vremenskih serija korištenjem radnog okvira Apache Spark." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. https://urn.nsk.hr/urn:nbn:hr:168:794093

Bošnjak, D. (2019). 'Prognoza vremenskih serija korištenjem radnog okvira Apache Spark', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 14 November 2024, https://urn.nsk.hr/urn:nbn:hr:168:794093

Bošnjak D. Prognoza vremenskih serija korištenjem radnog okvira Apache Spark [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2019 [cited 2024 November 14] Available at: https://urn.nsk.hr/urn:nbn:hr:168:794093

D. Bošnjak, "Prognoza vremenskih serija korištenjem radnog okvira Apache Spark", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2019. Available at: https://urn.nsk.hr/urn:nbn:hr:168:794093

Metadata
TitlePrognoza vremenskih serija korištenjem radnog okvira Apache Spark
Title (english)Time Series Forecasting with Apache Spark Framework
AuthorDario Bošnjak
MentorKrešimir Pripužić (mentor)
Committee memberKrešimir Pripužić (predsjednik povjerenstva)
Committee memberMarin Vuković (član povjerenstva)
Committee memberJurica Babić (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2019-07-12, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
Prognoziranje vremenske serije omogućuje uvid u budućnost promatranog procesa. Za uspješno prognoziranje najprije je potrebno provesti analizu vremenske serije koja obuhvaća dekompoziciju, statistička testiranja na stacionarnost te prikaz funkcija (parcijalne) autokorelacije. Prognoziranje je moguće raditi pomoću klasičnih statističkih metoda kao što su modeli AR, MA, ARIMA, eksponencijalno zaglađivanje i vektorska autoregresija. Uz klasične statističke metode moguće je koristiti i metode strojnog učenja, preciznije regresijske algoritme kao što su linearna regresija, model linearne regresije, slučajne šume i gradijentno ojačana stabla. Radni okvir Apache Spark za raspodijeljenu obradu i izračunavanja posjeduje funkcionalnosti pogodne za rad s vremenskim serijama te kroz knjižnicu za strojno učenje nudi navedene algoritme strojnog učenja s mogućnosti finog podešavanja postavljanjem hiperparametara. Studijski slučaj prognoziranja dnevne proizvodnje mlijeka pokazao je da su metode strojnog učenja ostvarene pomoću radnog okvira Apache Spark dale bolje rezultate od klasičnih statističkih metoda za prva tri koraka prognoziranja. Također se pokazalo da su meta-algoritmi ansambala dali bolje rezultate od linearne regresije.
Abstract (english)
Time series forecasting gives insights into the future behavior of the observed process. In order to make accurate forecasts, it is necessary to utilize time series decomposition, stationarity tests and to plot the (partial) autocorrelation function. Classical time series forecasting methods like AR, MA, ARIMA, exponential smoothing, and vector autoregression are widely used for time series forecasting. Besides classical methods, it is possible to use machine learning methods like linear regression, random forest, and gradient tree boosting. Apache Spark cluster-computing framework offers functionalities suitable for processing time series data and rich machine learning library. The library contains implementations of mentioned machine learning algorithms with the possibility of fine-tuning hyperparameters. Daily milk yield forecasting case study showed that machine learning methods implemented in Apache Spark produced better results than classical methods for the three-day-ahead forecast. The case study also showed that ensemble methods produced better results than linear regression algorithm.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:168:794093
Study programmeTitle: Computing
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2020-01-07 16:55:48