20.500.12556/DKUM-87212 Napovedovanje maloprodajnih cen mesa na podlagi časovnih vrst na študijskem programu 2. stopnje Matematika Predicting retail meat prices based on time series magistrsko delo V magistrskem delu se srečujemo z izzivom napovedovanja cen. Sprašujemo se, kateri so tisti parametri, ki vplivajo na končno ceno izdelka. Obravnavamo dve vrsti modelov napovedovanja cen - model multiple regresije, ki napoveduje ceno s pomočjo drugih parametrov, in sezonski ARIMA model, ki napoveduje cene na podlagi preteklega vzorca podatkov. Podatki, s katerimi gradimo modele, so v obliki časovnih vrst. V prvem delu uredimo podatke v skupno Excel tabelo ter izračunamo vse pomembne statistike. Sledi gradnja modela multiple regresije za posamezne izdelke, nato še gradnja sezonskega ARIMA modela. Modele multiple regresije gradimo v programu EViews, ARIMA modele pa v programu Python. Skozi analizo ugotovimo, da imamo v podatkih prisotno močno sezonsko komponento, zato podatke desezoniramo in zgradimo še model multiple regresije na desezoniranih podatkih. Končna ugotovitev kaže na to, da regresijski model na desezoniranih podatkih najbolje pojasnjuje variabilnost odvisne spremenljivke. In the master's thesis, we confront the challenge of price prediction. We inquire about the parameters that influence the final product price. We address two types of price prediction models - a multiple regression model that predicts the price using other parameters, and a seasonal ARIMA model that forecasts prices based on past data patterns. The data used to build the models are in the form of time series. In the first part, we organize the data into a comprehensive Excel table and calculate all relevant statistics. This is followed by the construction of a multiple regression model for individual products, and then the construction of a seasonal ARIMA model. We construct multiple regression models in the EViews software, while ARIMA models are built in Python. Through analysis, we discover a strong seasonal component in the data, so we deseasonalize the data and build a multiple regression model on deseasonalized data. The final conclusion indicates that the variability of the dependent variable is best explained by a regression model on deseasonalized data. Časovne vrste model multiple regresije ARIMA model sezonskost. Time series multiple regression model ARIMA model seasonality. true true false E. Kuhar Slovenski jezik Angleški jezik Magistrsko delo/naloga 2024-03-06 18:15:10 2024-03-26 13:21:14 2024-03-29 08:54:11 0000-00-00 00:00:00 2024 0 Maribor 0 Maribor VIII, 93 f. 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 519.237(043.2) 190132483 MAG_Kuhar_Eva_2024.pdf MAG_Kuhar_Eva_2024.pdf 1 E3797D91D9E44798C2E9553CDC16AB48 9f61a024c43fa3fd5685a51679184142dbf851abb8c92225a488017387e8ad06 50eb5e58-dbf5-11ee-bc18-00155d000105 https://dk.um.si/Dokument.php?lang=slv&id=176605 Fakulteta za naravoslovje in matematiko 0 0 0