Post-graduate theses

Search command : =""

Current Record: 8 of 145

[Add to Basket]
Identifier	000450288
Title	Comparative study of machine learning algorithms in economics's interest time series's data
Alternative Title	Συγκριτική μελέτη αλγόριθμων μηχανικής μάθησης σε οικονομικού ενδιαφέροντος δεδομένα χρονολογικών σειρών
Author	Αναγνωστάκης, Μιχαήλ
Thesis advisor	Τσιώτας Γεώργιος
Reviewer	Τζίνιους Μαργαρίτα Τσαγρής Μιχαήλ
Abstract	At the first part of this master thesis, initially we will explain neural net¬works and KNN, the theoretical background of applying k-nearest neighbor to time series forecasting. Afterwards we will make an application in monthly TOYOTA car sales (registrations) in Greece, data from AMVIR from Jan¬uary 2010 until and October 2021. We generate multiple KNN regression models, in order to make comparison. The simpler of them, with k=3, k=2, k=l, produce encouraging predictive results for our data set. In addition we generate an exponential smoothing model as a strict comparing basis for the kNN regression models.
Language	English
Subject	Forecasting
	KNNregression
	Machine learning
	ΚΝΝ παλινδρόμηση
	Μηχανική μάθηση
	Προβλέψεις
Issue date	2022-06-20
Collection	School/Department--School of Social Sciences--Department of Economics--Post-graduate theses
	Type of Work--Post-graduate theses
Permanent Link	https://elocus.lib.uoc.gr//dlib/8/3/b/metadata-dlib-1661851150-93872-10277.tkl
Views	394

Digital Documents
	Download document View document Views : 3