| dc.description.abstract | Demand forecasting is an important activity that is necessary for the Sales and Operation Planning
(S&OP), which is a popular integrated business management technique utilized by many
organizations. A case study of project between Suntory PepsiCo and YCH Protrade has been
addressed on several time transactions in which the challenge of projecting daily demand for
various product categories at the distribution level will be handled. The forecasting problem is
approached as a supervised machine learning task and evaluated by several metrics.
Based on insufficient data, the number of beverage items is projected in this article using the
Support Vector Regression (SVR), a method based on Support Vector Machine (SVM) algorithm,
and xGBoost method. Because SVR and xGBoost have several benefits, such as high generalization
performance and ensuring global minimum for given training data, it is expected that the predicting
system would perform well in anticipating customers’ demand.
It has also been shown that machine learning approaches not only give more accurate forecasts but
are also more suited for use in large-scale demand forecasting scenarios, which are common in the
FMCG industry. | en_US |
