Intermittent demand forecasting

• Main Author: Mishra, Prerna.
• Other Authors: Huang Guangbin
• Format: Final Year Project
• Language: English
• Published: 2013
• Subjects: DRNTU::Engineering::Manufacturing::Inventory control
• Online Access: http://hdl.handle.net/10356/54570
• Institution: Nanyang Technological University

Demand forecasting plays important role in synchronized planning. Business entity tries to understand the demand for a product for inventory scheduling. Such forecast holds tremendous importance to business because erroneous forecast could lead to extremities of „over supply‟ or „under supply‟- former leading to damage to business by glut in the market and later by loosing brand image by failing customer‟s expectation. Numerous models are employed to achieve forecast using historical information of demand. Forecast of demand of products having regular and uniform usage pattern is easily accomplished. But, it is extremely difficult to make precise forecast of the demand of products having intermittent demand pattern. Intermittent demand is one that is infrequent, occurs at irregular intervals. It refers to demand pattern for a product in which there are several periods in which there is zero and uneven demand. The modelling of forecast of intermittent demand is a challenging task requiring requisite modelling technique. The confidence on any technique is contingent upon the degree of accuracy of prediction for in-sample and out-of-sample forecasts. Literature is replete with several models proposed by researchers to resolve the problem of intermittent demand forecasting. However, comparison of result obtained for in-sample and out-of-sample forecasts by different modelling tools for same data of intermittent demand is scarce. The project paper deals with comparative assessment of the of forecast for “in-sample” as well as “out of sample” periods derived by various methods of intermittent demand forecasting on historical data set of fifteen hundred products for a span of twenty four months obtained by several numerical modelling tools – classical to contemporary and evaluate the superiority of models by comparison of achievement of different methods in respect of degree of accuracy of forecast for future periods and judge the robustness of forecasting methodology.