magistrsko delo
Aljaž Razpotnik (Author), Damjan Strnad (Mentor), Štefan Kohek (Co-mentor)

Abstract

V magistrskem delu se ukvarjamo z napovedovanjem časovnih vrst. Časovno vrsto predstavljajo podatki vremenskih parametrov večjega števila krajev, ki smo jih pridobili iz spletnega arhiva Agencije Republike Slovenije za okolje. Za napovedovanje vremenskih parametrov izbranega kraja uporabimo pretekle podatke samega kraja in njegove okolice ter z njimi učimo napovedne modele ARIMAX, CART, GRU in kombinirani model CNN-LSTM. Pri kombiniranem modelu upoštevamo geografske soodvisnosti uporabljenih krajev, ki jih preslikamo v matriko. Iz predstavljenih rezultatov je razvidno, da sta najboljša modela za napovedovanje vremenskih parametrov GRU in CNN-LSTM.

Keywords

vremenski parametri;časovna vrsta;napovedovanje;nevronska mreža;regresijsko drevo;magistrske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: [A. Razpotnik]
UDC: [519.2:004.8]:551.509(043.2)
COBISS: 22877718 Link will open in a new window
Views: 928
Downloads: 159
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Short-Term prediction of local weather parameters with convolutional neural network
Secondary abstract: The master thesis deals with time series forecasting. The time series is presented by the data of weather parameters of a large number of places, which are obtained from the web archive of the Slovenian Environment Agency. In order to forecast the weather parameters of a specific place, past data of that place and its environment are used for teaching the forecast models ARIMAX, CART, GRU and the combined CNN-LSTM model. The combined model requires the consideration of geographic interdependencies between places, which are mapped into the matrix. The results demonstrate that the most effective models for forecasting the weather parameters are GRU and CNN-LSTM.
Secondary keywords: weather parameters;time series;forecasting;neural network regression tree;
Type (COBISS): Master's thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije
Pages: X, 53 str.
ID: 11279858