Raw materials forecasting methods of an outsourced ready-to-wear suit factory

Abstract

The objective of this paper is to design raw materials forecasting methods for material management process
improvement of an outsourced ready-to-wear suit factory. In this research, we study demands of five types of
frequently used raw materials and identify the most appropriate forecasting method providing the lowest
Mean Absolute Percentage Error (MAPE). We found that Simple Exponential Smoothing method gives the
most accurate result with smoothing parameters (α ) of 0.9439, 0.6906, 0.8656, 0.9844 and 0.8694 for SKU number EX102, RX101, HX101, IX103 and KX101, respectively. From the results, we found considering
weekly demand yields more accurate forecasting results as compared to monthly demand. From our on-site
experiment from October to December 2012 to identify the suitable time to adjust smoothing constant (α ),
we found that the smoothing constant should be reviewed and adjusted every 4 weeks for SKU number
RX101, HX101 and KX101, giving the MAPE results of 21.125, 13.170 and 18.952, respectively. For EX102
and IX103, the parameter should be reviewed every 8 weeks, giving the MAPE results of 7.538 and 8.897,
respectively. Overall, the proposed model can reduce forecasting errors up to 45 percent as compared to the
naïve method previously used in this factory.