Enhancement of short-term forecasts : A study of a pharmaceutical distributor

Sammanfattning: Forecasts are used to predict the uncertain outcome of a variable. These predictions are made to get an understanding of likely future scenarios which allows planning in advance. Forecasts are commonly used in inventory control systems to estimate the future demand. This estimation along with the estimated precision of the forecast can be used to determine adequate safety stocks. In extension, forecasts of the future demand can be used to support decisions regarding the replenishment of inventories.     This thesis involves forecasting of the demand of pharmaceutical drugs. In particular, drugs that are referred to as generics. The demand of generics can fluctuate heavily due to the substitution system and can therefore be troublesome to forecast. The main objective of this thesis is to design a suitable forecast approach for the demand of generics, and to examine how these forecasts can be used to control the replenishment of inventories. The project was executed at a pharmaceutical drug distributor and hopes to enlighten some new techniques that can be used to improve the handling of generics. To achieve the objective of this thesis, the initial focus was on the current state. The currently used methods where studied and compared with theories and methods described in literature. Assumptions and theories on which the currently used methods are based on could thus be identified. These assumptions where later assessed as either reasonable or unreasonable. Finally, a new forecast approach was designed to account for the insights gained from evaluating the current methods. The result of this thesis is a forecast approach suitable to forecast the demand of generics that account for the fluctuation in demand which occurs due to the substitution system. It assumes that the demand has a constant demand model, but an alternative method suitable for the trend model is also given. It is proposed that tracking signals are used to monitor the forecast such that systematic errors are identified. Furthermore, the literature review indicates that alternatives to the current method used to determine the reorder point should be considered for occasionally demanded generics.