20.500.12556/DKUM-86158-1c270c83-430a-d5cf-602b-03fd2525ac4e Napovedovanje gostote prometa z uporabo strojnega učenja Traffic density forecasting using machine learning V magistrskem delu naslavljamo problem dolgoročnih napovedi prometa. V ta namen najprej predstavimo sorodno delo in podamo teoretično osnovo izvedbe različnih modelov strojnega učenja ter kodiranja cikličnih podatkov. V nadaljevanju podrobneje predstavimo naš pristop, ki omogoča izvedbo letnih napovedi z urno ločljivostjo. Različne pristope pri tem sistematično primerjamo in z rezultati pokažemo, da smo obravnavni problem najučinkoviteje naslovili z uporabo metode XBoost in kodiranja cikličnih podatkov s podobnostjo. In this master’s thesis, we address the problem of long-term traffic forecasts. For this purpose, we first address related work and provide the theoretical basis for the implementation of various machine learning models and representations of cyclic data. In continuation, we detail the proposed approach that allows for annual forecasting with an hourly temporal resolution. The different machine learning models used are than systematically compared. As shown by the results, the best results for the addressed problem were obtained using the XBoost method and encoding cyclic data with similarity. strojno učenje promet ansambel dreves značilnice obdelava podatkov machine learning traffic ensemble trees features data processing true true false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2023-10-15 12:40:05 2023-11-07 07:10:43 2023-11-08 13:43:44 0000-00-00 00:00:00 2023 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 MAG_Kralj_Andraz_2023.pdf MAG_Kralj_Andraz_2023.pdf 1 E0222F147E3F8D3A3D0ACBACDF688D05 0bb47702ed2e9f5657168f6774826af03a3b636c7f44df8f5be39327b3842222 7820f15c-7cd6-11ee-96c9-00155d000105 https://dk.um.si/Dokument.php?lang=slv&id=173975 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0