LAUSR

In several industries, global competition, increasing customer expectations and technological innovations tend to accelerate product life-cycles. In this changing environment, traditional forecasting methods tend to be ineffective as a consequence of the transient and highly uncertain demand of short life-cycle products (SLCP), and the scarcity of sales data. To address this challenge, we present a methodology to forecast SLCP demand using time series of similar products referred to as analogies. Linear regression and clustering techniques are used for the selection and weighting of suitable analogies. The proposed methodology is tested against seven analogue-based forecasting methods, including two implementations of non-linear regression methods. In different sets of time series, our methodology attained more accurate forecasts with short processing times compared with state-of-the-art methods. Such results reveal promising applications of combined regression and clustering techniques as simple and effective forecasting tools for supporting replenishment decisions for SLCP.