<![CDATA[The competitive animal feed industry requires efficient production planning to meet market demand, maximize resource use, and sustain profitability. Various raw materials, tools, and techniques are utilized to create animal feed, which results in various variants that might influence throughput rates and thereby alter the accuracy of yield projections. Data mining is applied to train and validate different algorithms to ascertain the most effective model for predicting throughput rates through machine learning. This study uses CRISP-DM to construct an enhanced predictive model for production throughput rate. Due to the model's improved prediction accuracy, scheduling and operational decision-making will be more efficient and cost-effective. The CRISP-DM framework is used to examine historical production data and forecast production levels. Advanced machine learning techniques train and evaluate the model to make accurate predictions that can be mathematically simulated using possible constraints. The findings show that throughput rate predictions are effectively generated by the predictive model that was created using data mining processes. Metrics like Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) are used to assess the model and identify the optimal model after attempting using different predictive machine learning techniques. With the linear regression algorithm and MAE values of 5,186, MSE of 1,585, and RMSE of 5,970.32, the best prediction model test results have been determined. An optimal scheduling simulation is conducted from the selected model, with the constraint of the customer's delivery requirements and the time capacity, specifically a maximum throughput rate prediction of 23.78 tons/hour. However, this study reveals how the data mining process is applied to the decision-making process with the use of operation research support so that the optimal production rate prediction is 22 tons/hour.]]>
Copyrights © 2025
