[ Log In | Sign Up | Lost password? ]

Forecasting in manufacturing

Forecasting is a general need in real life and in business. A general effective forecasting method doesn't exist and in every application field you should look for characteristics of the problem that could make your forecasting activity easier and more effective. In a manufacturing environment you must start clearifying why you need forecasts. The main reason should follow from the fact that your customer accepts some delivery lead times while you need more time than that to purchase raw materials and components and to manufacture and assemble the finished products. Being so, you have to anticipate some activities before you have acquired the customer order. And to anticipate effectively these activities (could be purchasing some raw materials or manufacture some semifinished products: let's call them time-critical items) you need to forecast customer orders and translate them in time-critical items' needs.

Marketing forecasting becomes easier

In many companies the marketing department forecasts sales for single items or orders aggregations like sales areas, salesman, time bucket. This is right for understanding how a product is appreciated or an agent is good at selling or a market is subjected to seasonality but it doesn't help the production planning department to prepare the time-critical items and, indirectly, to fulfill the customer orders effectively. It would be helpful to provide the production planning department with a differently aggregated forecast. This kind of 'forecast for manufacturing' should consider the commonality of time-critical items in bill of materials.

Planning bills

But marketing departments usually don't mess with bill of materials, so the newly aggregated forecast should be done by someone else (MIS? planning department power users?) and the should be done by means of planning bills. These are abstract bill of materials, whose structure is specific to the industry considered and that can be created in databases using clustering algorithms.