Demand & Inventory Management
The forecast model provided by the overseas office Is often inaccurate because the forecasting is performed before the actual production schedule and It is based on marketing survey results and historical data from an overseas research team. This means that the forecast provided by the overseas office has a long lead time and It Is not accurate because of dynamic changes such as the global economic conditions and customer taste. In real life, demand is characterized by dynamic structural change.
This paper investigates the problem of an inventory control system in a high technology batch production environment. The major characteristics of the problem are: . There is some standardization of products and there are repeated orders for these standard products.. There Is some specialized customization of the products, but this forms a small portion of the total workload of the company.. The market uncertainty faced by the batch production environment are: the quantity needed to be produced; and the timing of the customer orders..
There Is a lead time between a placement of a purchase order of parts and materials from outside supplier and the delivery date of the parts and materials. The lead time of the parts and materials can be
Need essay sample on "Demand & Inventory Management"? We will write a custom essay sample specifically for you for only $ 13.90/page
A case study is then presented to illustrate the application of the model based on data from a multinational company in Singapore. It also compares the Bayesian dynamic linear time series model with a lassie forecasting model (autoregressive integrated moving average (IRMA) model). The limitations of the existing forecasting and inventory management techniques are: . They cannot cope with the dynamic or structural change of the time series data.. They fail to characterize the inventory problem with limited historical data (fewer than 100) (Schillings, 1988)..
They do not adequately address issues relating to the short life cycle and nonsensical products with relatively long lead time. The concepts of economic order quantity (EX.) and Parent analysis provide the basis for lancing the cost associated with inventory replenishment decisions (Buff and Miller, 1979; Buff and Saris, 1987; Adam and Bert, 1986; Evans et al. , 1987). However, the conventional deterministic inventory system and standard statistical approaches are unable to overcome the above mentioned limitations (Buff and Miller, 1979; Levin et al. , 1972; Peterson and Silver, 1979).
The limitations of the existing forecasting techniques on the independent demand items are very well documented (Tailored, 1989; Evans et al. , 1987; Saying and Kirkpatrick, 1994). Bayesian dynamics time series and forecasting techniques can be used to solve inventory problems because Bayesian inference statistics has the analogue idea that posterior knowledge (actual sales demand) can be derived from prior knowledge (such as the manager’s experience) and the likelihood (the similar or expected trend) of the product demand (Box and Tioga, 1973; Jeffrey, 1961; Lee, 1988; Press, 1989).
In many real life forecasting problems (for example when previous demand data are not available for newly launched products), there is little or no useful information This work was carried out while the author was Associate Professor in the School of Mechanical and Production Engineering at Nanning Technical University in Singapore. Integrated Manufacturing Systems 11/5  331В±339 # MAC University Press [SINS 0957-6061] [331 ] T. A. Speeding and K. K. Chain Forecasting demand and inventory management using available at the time when the initial forecast is required.
Hence, the early forecast must be based largely on subjective considerations (such as the manager’s experience and the general demand of a similar or comparable product). As the latest information (actual sales demand) becomes available, the forecasting model is edified with the subjective estimation in the presence of the actual data. This paper compares the Bayesian dynamic linear time series forecasting method to auto- regressive integrated moving average (IRMA) analysis (Box and Jenkins, 1976) for the purposes of demand forecasting.
To illustrate the demand forecasting techniques, a case study of a manufacturing company is presented in the paper. Three different forecasting models with three different sets of data (the first 24 weeks, the first 36 weeks and the complete 51 weeks data) are built using Bayesian dynamics time series and forecasting and IRMA techniques. The three different sets of data are used to investigate the dynamic behavior and structural change of the system. From the modeling results, a comparative study of the models’ accuracy is performed.
The strengths and weaknesses of the Bayesian time series analysis (BATS) models are also discussed and compared to the IRMA model. The Bayesian dynamic linear time series model Traditional forecasting approaches are based on characteristic the structure of historical time series and then predicts future events based on that structure. Obviously, the structure of the time series may change in a volatile business environment. The parameters of the time series model would then need to be re- estimated based on the new structure of the time series.
Bayesian Forecasting, however, is based on the principle that routine forecasts can be updated by subjective intervention as external information becomes available. In this context ‘ ‘Bayesian learning” is particularly relevant to applications in a dynamically changing environment. Bayesian dynamic time series and forecasting is the core theory adopted by BATS. BATS is a computer software developed by Pole et al. (1994). BATS is designed for PC-compatible machines and is available in DOS and Windows versions.
BATS is menu driven and provides all the modeling facilities discussed and exemplified in the reference book (Pole et al. , 1994) (see Figure 1). BATS has been successfully used to solve time series problems in social and economic research (Pole et al. , 1994). BATS provides model building functions (response, trend, seasonal, regression and variance), forecasting monitoring functions (Bases’ theorem), intervention analysis tools (subjective input from the user), error analysis capability (time plot of the forecasting residuals) and graphical visualization tools.
The sequential modeling procedure of BATS is as follows: 1 Preliminary investigation and model identification. The time plot of the time series is investigated at this stage. This is to identify the trend cyclic, seasonal, stationary and missing values of the data. 2 Model building. The model is built by using the BATS software. At first, the prior information is initialized based on expert knowledge. In the case study presented in this paper, the expert knowledge is based on a monitoring scheme of the BATS software is used to monitor the changing trends so that the user can input the appropriate invention..
This is important for identifying tutorial change, which is typical in a dynamic environment, such forecasting demand for logistical management. 4 Accuracy checking. The residuals of the forecasted time series, auto-correlation (ACE) and partial auto-correlation function (PACE) of the residuals are plotted to check if they are significantly different from zero. This analysis verifies the accuracy of the model. 5 Forecasting. The forecasted values can be generated by the BATS software with the Bayesian formula automatically.