Višestruka regresija i prognoza ukupne akcijske prodaje

master's thesis

• Author: Nina Šunjo
• Mentor: Miljenko Marušić (mentor)
• Committee member: Miljenko Marušić (predsjednik povjerenstva)
• Committee member: Matija Kazalicki (član povjerenstva)
• Committee member: Marko Vrdoljak (član povjerenstva)
• Committee member: Pavle Goldstein (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Mathematics); Zagreb
• Defense date and country: 2014-09-23, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Mathematics

Abstract

Cilj ovog istraživanja bio je upoznati se s koristima grupiranja podataka u razvitku matematičkog modela za predviđanje akcijske prodaje, te razvitak tog modela. Za grupiranje podataka korišten je k-means algoritam, a metodom višestruke linearne regresije vršene su procjene. Regresijom je izrađen model uz korištenje stvarnih podataka dobivenih od trgovačkog lanca Konzum. Rezultati su pokazali da grupiranje dobivenih podataka na pet grupa ne smanjuje grešku predviđanja, čak je povećava. Mijenjanjem raznih parametara modeliranja te korištenjem drugih vrsta podataka možda bi se i došlo do boljih rezultata, stoga je potrebno daljnje istraživanje u svrhu poboljšanja postojećeg modela.

Abstract (english)

The aim of this study was to become familiar with the benefits of grouping data in the development of a mathematical model to predict the action of sales, and the development of the model. K-means algorithm was used to group the data, and the method of multiple linear regression was used to perform evaluation. Regression model was created using actual data provided by the retail chain Konzum. The results showed that clustering data into five groups does not reduce the prediction error, but increases. By changing various parameters of modeling and the use of other types of data might achieve better results, so further research is needed to improve the existing model.

• Keywords: razvitak matematičkog modela za predviđanje akcijske prodaje; k-means algoritam; metoda višestruke linearne regresije
• Keywords (english): development of a mathematical model to predict the action of sales; K-means algorithm; multiple linear regression model
• Language: croatian
• URN:NBN: urn:nbn:hr:217:954184
• Study programme: Title: Mathematical Statistics; 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: 2019-02-14 11:22:56