Demand forecasting by machine learning: (case study of acrylic products)

In the world of business with many strong competitors, demand forecasting isone main key in the supply chain. Demand forecasting with good accuracy duringuncertainty is very challenging. Traditional forecasting is not capable when there areuncertainties. In such a case, the machine learning method is proposed to predict salesof acrylic products which is time series data through the uncertainty of the COVID-19.The last seven years’ sales data is provided by the acrylic company in Thailand. Theforecasted sale from proposed Artificial Neuron Networks method is compared withthe traditional Holt’s Double Exponential Smoothing forecasting methods bycomparing the amount of prediction error to determine the prediction accuracy. Finally,the best suitable model combination is proposed for all products properly.