Primjena statističkih metoda u predviđanju prodaje

master's thesis

• Author: Karla Šutalo
• Mentor: Siniša Slijepčević (mentor)
• Committee member: Siniša Slijepčević (predsjednik povjerenstva)
• Committee member: Ilja Gogić (član povjerenstva)
• Committee member: Igor Pažanin (član povjerenstva)
• Committee member: Zvonimir Tutek (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Mathematics); Zagreb
• Defense date and country: 2017-09-28, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Mathematics

Abstract

Cilj ovog rada je naći model koji najbolje opisuje podatke o prodaji grupe proizvoda i napraviti predviđanje buduće prodaje. Podaci o prodaji su vremenski niz za koji definiramo najvažnije pojmove: stacionarnost, autokovarijacijsku i autokorelacijsku funkciju. Kod modeliranja podataka koristimo teoriju vezanu uz stacionarne procese pa opisujemo metode uklanjanja trenda i sezonalnosti iz podataka kako bi dobili stacionaran niz koji dalje modeliramo. Definiramo ARMA procese, odnosno autoregresivne procese pomičnih prosjeka i navodimo neka njihova važna svojstva. Obrađenu teoriju primjenjujemo na stvarnim podacima o prodaji grupe proizvoda, gdje dio podataka modeliramo ARMA procesom i predviđamo njihovu prodaju. Na kraju dobiveno predviđanje uspoređujemo sa stvarnim podacima o prodaji i zaključujemo kako ARMA proces daje dosta dobra predviđanja prodaje.

Abstract (english)

The goal of this thesis was to develop a model which describes the sales data for the group of three products and predicts group’s future sales. Sales data is time series for which we define the most important terms: stationarity followed by autocovariance and autocorrelation function. Using theoretical knowledge we defined statistical methods for removing trend and seasonal component from the time series which then leaves us with stationary time series used for modeling. We defined ARMA process, which stands for the Autoregressive-moving-average process, and define some of its properties. We applied theoretical knowledge on actual sales data for the product group by developing an ARMA model on data sample and predicting future sales. Finally, we compared predicted future sales with realized sales and came to the conclusion that ARMA process describes our data well and gives good sales prediction.

• Keywords: predviđanje prodaje; modeliranje podataka; ARMA proces
• Keywords (english): sale prediction; dana modeling; ARMA process
• Language: croatian
• URN:NBN: urn:nbn:hr:217:189000
• 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: 2018-04-27 13:14:07