Analiza podatkov z metodo delnih najmanjših kvadratov (PLS metoda)

delo diplomskega seminarja

Žiga Trojer (Author), Marjetka Krajnc (Mentor)

Abstract

V delu diplomskega seminarja definiramo linearne modele več spremenljivk v splošnem. Nato predstavimo model po metodi najmanjših kvadratov, kjer si ogledamo glavne značnilnosti modela in izpostavimo glavne pomankljivosti za različne tipe podatkov. Nadaljujemo s predstavitvijo sorodnega modela regresije glavnih komponent, kjer predstavimo glavne ideje metode glavnih komponent. Ko smo seznanjeni z delovanjem te metode, podoben princip uporabimo na metodi delnih najmanjših kvadratov. Teorijo modela podkrepimo s primeri delovanja metode v napovedovanju vrednosti spremenljivk. Nato se dotaknemo še problema klasifikacije in nelinearnih modelov, kjer delovanje prikažemo z enostavnima zgledoma.

Keywords

matematika;delni najmanjši kvadrati;NIPALS;regresija;klasifikacija;redukcija dimenzij;

Data

Language:	Slovenian
Year of publishing:	2019
Typology:	2.11 - Undergraduate Thesis
Organization:	UL FMF - Faculty of Mathematics and Physics
Publisher:	[Ž. Trojer]
UDC:	519.2
COBISS:	18741081
Views:	2123
Downloads:	399
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Data Analysis Using the Partial Least Squares Method (PLS method)
Secondary abstract:	In the seminar, we define linear models of several variables in general. Then we introduce the model using the least squares method, where we look at the main features of the model and highlight the main disadvantages for different types of data. We continue to introduce a related model of principal component regression, where we present the main ideas of the principal component method. After we are familiar with how this method works, we apply a similar principle to the method of partial least squares. The theory of the model is supported by examples of how the method works in prediction. Finally we look also at the problem of classification and nonlinear models, where we show the operation on some simple examples.
Secondary keywords:	mathematics;partial least squares;regression;classification;dimensionality reduction;
Type (COBISS):	Final seminar paper
Study programme:	0
Embargo end date (OpenAIRE):	1970-01-01
Thesis comment:	Univ. v Ljubljani, Fak. za matematiko in fiziko, Oddelek za matematiko, Finančna matematika - 1. stopnja
Pages:	27 str.
ID:	11221906