Inventory Control by Linear and Non Linear Demand Forecasting


  • R. V. Patil PG Student, Department of Mechanical Engineering, AMGOI Vathar, Maharashtra, India
  • A. N. Chapgaon Head of the Department, Department of Mechanical Engineering, AMGOI Vathar, Maharashtra, India


Decision support system (DSS)


Now a day, supply chain practices are widely adopted in Indian industries .Research points out examination of success factors and implementations of the system in Indian industries. However, the adoption in Small and Medium Enterprises is not very common. Interestingly, multinational firms and large enterprises can invest huge capital for implementing latest information technology tools to share the information and carry day-to-day operations, but the investment and implementation is quite difficult for SMEs. This inspires us to investigate effect of new age supply chain technology like VMI practices in SMEs and other industries. VMI entails forecasting demand through joint efforts of customer and supplier, maintaining a targeted service level for customers, initiating and shipping supply orders, material control and customer order fulfillment.In this study, the results of adopting a partial vendor managed inventory practice, along with latest decision support tool like ANN, are presented. Outcomes of case study shows that deployment of vendor managed forecasting improves forecasting accuracy, reduces bullwhip, minimizes total supply chain cost, improves profits and most importantly improves customer satisfaction indexOverall five statistical models and five neural network models are adopted and compared. Study illustrates how a neural network aptly learns the case dynamics, and improves system performance. The results presented in this section demonstrates the effectiveness of the Focused Time Lagged Recurrent Neural Networks (FTLRNN) model compared to traditional and other neural network models. The significant finding of this research is results of forecasting error and other supply chain performance measures. Further study reveals that when we bracket the overstock and under stock cost in the supply chain cost, a forecast with minimum forecasting error may not lead to reduced supply chain cost or improved profits. This study also introduces a mixed model where the error obtained from statistical model is mixed with the forecast obtained by neural model and a new forecast is obtained. The analysis shows that the developed model could further improve supply chain performance in VMI setting.