Answers to Your Forecasting Questions

answers to your forecasting questions q. I wonder if you can share with us your experience about the advantages and disadvantages of weekly forecasting. Which techniques should we use for best results? My company is a leader in the pharmaceutical products with 10,000 SKUs. We forecast every week, and 16 weeks into the future. Recently we changed our forecasting system. With that our forecasting error went up instead of going down. How can we improve the forecast error? Which process and model would you suggest? Teodoro Perez—Phama Products Distributor a. I’ll answer your questions one by one. 1. Forecast error increases as we forecast at a more and more granular level. The forecast errors of weekly forecasts are likely to be higher than of monthly forecasts. You may like to prepare first monthly forecasts and then break them down into weekly buckets. 2. There is no magic model I can suggest. Most businesses use Time Series models for short-term forecasting, which are simple and easy to use. Within Time Series, the models used most are averages and trend. You can develop ex post forecasts (preparing forecasts of those periods for which actuals are known), then determine which ...