CVS order quantity optimization through testing forecast models considering supplier reliability

Varied time-series models, and the associative model integrating the effect of purchase price to order quantity, considering monthly and quarterly data with the consideration of 91.67% supplier reliability, are evaluated in terms of forecast errors and losses in order to identify which is appropriate to project demands for catalyzed vinyl sealer (CVS) for a Cebu-based furniture company. Associative models are based on interrelating material price and actual material usage (AMU) level. On the other hand, time-series models are based on the patterns of the 3-year AMU levels. The accuracy of the 186 developed models is tested by the interpolated mean measures of errors MAD, MAPE and MSE. The analysis on the time value of money of the losses, of understocking and understocking of the top five models with the least interpolated errors, suggests implementing seasonal decomposition with forecasting deseasonalized data using exponential smoothing at alpha 0.20.