Metode linearne regresije i primjena u marketingu

master's thesis

• Title (english): Linear regression methods and application in marketing
• Author: Ana Ivković
• Mentor: Siniša Slijepčević (mentor)
• Committee member: Siniša Slijepčević (predsjednik povjerenstva)
• Committee member: Nikola Sandrić (član povjerenstva)
• Committee member: Goran Muić (član povjerenstva)
• Committee member: Nela Bosner (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Mathematics); Zagreb
• Defense date and country: 2022-09-28, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Mathematics

Abstract

U ovom diplomskom radu proučili smo linearnu regresiju s naglaskom na jednostavnu i višestruku regresiju. Za početak, rekli smo nešto o povijesti regresije, bivarijantnoj normalnoj distribuciji te njenim svojstvima. Proučili smo metodu maksimalne vjerodostojnosti te metodu najmanjih kvadrata, koja je i najstarija metoda, te je radi svoje jednostavnosti najčešće korištena za procjenu parametara. Od interesa nam je bila i analiza varijance te smo kroz primjer vidjeli i njenu primjenu. Kraj ovog diplomskog rada ilustrira primjenu suvremenih pristupa na skupu podataka tvrtke Nepa. Cilj posljednjeg poglavlja bio je konstruirati strategiju kako napraviti model prikladan za visoku razinu složenosti podataka. Marketing Mix model bavi se svim elementima problema koji se proučava. Jedan od tih elemenata je izbor odgovarajuće funkcionalne forme. U ovom slučaju, dijagram reziduala sugerira da je log-linearan model najprikladniji.

Abstract (english)

In this thesis, we are introduced to the linear regression with an emphasis on simple and multiple regression. To begin with, we said something about the origins of regression, the bivariate normal distribution and its facts. We have seen maximum likelihood method and the method of least squares, which is also the oldest method and, due to its simplicity, is most often used for parameter estimation. The analysis of variance was also of interest to us, and we saw its purpose of using through an example. The end of this thesis illustrates an application of modern approaches of statistical learning on a set of data provided by Nepa. The goal was to construct a model building strategy suitable for a high level of complexity of the data. A Marketing Mix model must address all elements of the problem being studied. One of such elements is the choice of the appropriate functional form. At the end, the plot of the residuals against each predictor variable suggest that the log-linear specification is appropriate in this case.

• Keywords: bivarijantnaj normalna distribucija; metoda maksimalne vjerodostojnosti; metoda najmanjih kvadrata; analiza varijance; tvrtka Nepa; Marketing Mix model; log-linearan model
• Keywords (english): bivariate normal distribution; maximum likelihood method; method of least squares analysis of variance; Nepa; Marketing Mix model; log-linear specification
• Language: croatian
• URN:NBN: urn:nbn:hr:217:503455
• Study programme: Title: Finance and Business Mathematics; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra matematike (magistar/magistra matematike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2022-10-26 12:07:31