| Abstrakt: |
The Bullwhip effect is a phenomenon that tends to increase the volatility in the demand distribution when moving upstream in the supply chain. Because of this, most of the industries are failing to accurately forecast the demand. This leads to increased manufacturing cost, higher inventory level, longer replenishment lead times, low product availability, higher transportation cost and overall, it reduces the supply chain profitability. Previous studies have proposed different methods to overcome the Bullwhip effect such as improving demand forecasting, reducing order batching, reducing the incentives of forward buying, better sharing of information, shorten lead and review period times and designing single-stage replenishment control. This study focuses on the customer-centric approach of categorizing the customers by incorporating Machine Learning techniques for better understanding the customers' purchasing behavior. Specifically, this study analyzes the hidden patterns and similarities that exist between different customers and the products that they purchase. For the analysis, Principal Component Analysis (PCA) is used for dimensionality reduction and k-Means Clustering, an unsupervised learning technique is used for customer categorization. By this, the end customer demands are tracked downstream from upstream in the supply chain, which may help to reduce the demand variation. [ABSTRACT FROM AUTHOR] |
