Božić D., Stanković R., Kolarić G Analysis of the pattern aggregation impact on the demand forecasting ISSN 1846-6168 UDK 62 ANALIZA UTJECAJA AGREGACIJE UZORAKA NA PROGNOZU POTRAŽNJE ANALYSIS OF THE PATTERN AGGREGATION IMPACT ON THE DEMAND FORECASTING Diana Božić, Ratko Stanković, Goran Kolarić Pregledni rad Sažetak: Opskrbni lanci su vrlo rijetko u jednostavnom generičkom obliku, već uključuju različite sudionike, koji neovisno provode prognozu potražnje u svojem djelokrugu. Prognoziranje potražnje temeljem podataka o narudžbama umjesto podataka o potražnji krajnjeg kupca (korisnika) u lancu, na sljedećim višim razinama kumulativno generira sve veća odstupanja. Svaki od sudionika opskrbnog lanca tako dobiva drugačije podatke o potražnji što je uzrokovano tim kumulativnim djelovanjem, poznatim pod nazivom “efekt biča”. U cilju smanjivanja ovih nepravilnosti, proizvođači obvezuju distributere na dostavljanje podataka o prognozi potražnji na svojem tržištu. Distributeri su pritom suočeni s problemom prikupljanja i obrade heterogenih uzoraka potražnje od ostalih sudionika na nižim razinama. U radu je prikazana analiza agregacije uzoraka koji su korišteni za prognoziranje potražnje primjenom različitih metoda prognoziranja. Ključne riječi: metode prognoziranja na vremenskim nizovima, agregacija uzoraka potražnje, efekt biča, opskrbni lanac Review article Abstract: Supply chains are rarely in their basic, simple form – they involve different participants who respectively use demand forecasting methods related to their filed. Demand forecasting based on orders received instead on end user demand data will inherently become more and more inaccurate as it moves up the supply chain. Each participant in a supply chain receives different fluctuations data in the orders obtained, which is caused by the bullwhip effect. In order to mitigate these distortions, producers require the distributors to deliver the data on demand forecasting for a certain market. Thus the distributor tries to find the appropriate forecast method. This can be very difficult since the demand patterns of buyers differ. The paper analyses the pattern aggregation used for demand forecasting by applying different forecasting methods. Key words: time series forecasting methods, demand pattern aggregation, bullwhip effect, supply chain 1. INTRODUCTION In the basic form, a supply chain consists of a company with its suppliers and customers [1]. Extended supply chains have more participants, such as supplier's supplier, service providers, and customer's customer. Upwards and downwards the supply chain structure, a significant discrepancy in customer demand information between different stages occurs. This phenomenon is known as bullwhip effect or Forrester or whiplash effect [2],[3]. Producers and other participants want to avoid disturbances in their business plans, especially when dealing with short shelf life inventory. For this reason, the right order quantity must be determined, to meet the business plan the best way. This can be achieved by implementing adequate forecasting method and to periodically verify if the method yields the expected results with reference to the respective demand pattern. The main problem involved is aggregation of demand patterns acquired from different sources. This paper deals with different time series forecasting methods, and models of demand patterns aggregation used by the respective method. 2. MITIGATING DISTORTION IN SUPPLY CHAIN Managing supply chain requires trade-offs between efficiency and effectiveness of the participants involved. This can be seen in logistics operations planning which among other issues deal with maintaining the right balance among production, inventory and distribution [4]. Those decisions are based on forecasts that define which products will be required, what amount of these products will be called for, and when they will be needed. Demand forecasting becomes the basis for mitigating distortions and is used by companies to plan their internal operations and to cooperate among each other to meet market demand. All forecasts deal with four major variables: supply, demand, product characteristics and competitive environment. They combine to determine 426 Technical journal 7, 4(2013), 426-430