Decision support tools for sales forecasting

Abstract

CPRS Semiconductor Corporation is an Chinese multinational corporation. It designs and manufactures semiconductors products like Random Access Memory (Commonly known as RAM), Universal Serial Bus (Commonly known as USB), capacitive touch-sensing controllers (used in smart devices like smartphones, tablets, laptop screens etc.) and several types of Integrated Chips (Commonly known as ICs). The company’s offerings cater to the needs of various industries like automotive, handsets, telecommunications etc.. One of the major challenges the firm is facing is forecasting the demand of its product offerings. This challenge is omnipresent in the whole industry2 and thus capacity planning, manufacturing decision of whether to outsource or produce in-house, manpower planning all become more difficult to carry out. CPRS is facing similar problems at present and they also planned to use better forecasting models to carry out decision making more efficiently. CPRS has its own forecasting methods presently which gets input from its sales department. The project work required to come up, independently, with a forecasting techniques which can better fit the requirement of the firm. Forecasting methods which are mostly used like moving average, regression, exponential smoothing were tried without much accuracy. Auto Regression Moving Average Method (ARMA) which combines auto regression (based on previous quarter’s data) and moving average of errors in forecasting in the previous quarters was also used. Surprisingly this method didn’t yield satisfactory results. Finally some combination of methods mentioned above were tried and combination of regression and auto-regression methods gave us a satisfactory results given the context that all the well-known methods were not yielding any result. At this point we are not aware of this method being used in any research paper but we expect it must have been used somewhere.