Demand of beer is uncertain and has seasonal fluctuation. Over the past years, based on demand forecast and quantity of semi-finished product to assess the input amount of materials, and have not evaluated the optimal production time, so it was difficult to balance the demand and supply. This research is based on historical data to infer the demand distribution, and applied statistical cluster analysis technique to classify the off-season and season demand. We applied the optimization technique to search for an optimal production quantity to maximize the total expected profit. We also establish a mixed-integer linear programming model for production scheduling to minimize the total production losses according to the given production quantity per week. Therefore, the quality of production scheduling was improved. Finally, this paper takes a real case to demonstrate the validity of the proposed model.