This study applies response surface methodology (RSM) to the hyperparameter fine-tuning of three machine learning (ML) algorithms: artificial neural network (ANN), support vector machine (SVM), and deep belief network (DBN). The purpose is to demonstrate RSM effectiveness in maintaining ML algorithm performance while reducing the number of runs required to reach effective hyperparameter settings in comparison with the commonly used grid search (GS). The ML algorithms are applied to a case study dataset from a food producer in Thailand. The objective is to predict a raw material quality measured on a numerical scale. K-fold cross-validation is performed to ensure that the ML algorithm performance is robust to the data partitioning process in the training, validation, and testing sets. The mean absolute error (MAE) of the validation set is used as the prediction accuracy measurement. The reliability of the hyperparameter values from GS and RSM is evaluated using confirmation runs. Statistical analysis shows that (1) the prediction accuracy of the three ML algorithms tuned by GS and RSM is similar, (2) hyperparameter settings from GS are 80 reliable for ANN and DBN, and settings from RSM are 90 and 100 reliable for ANN and DBN, respectively, and (3) savings in the number of runs required by RSM over GS are 97.79, 97.81, and 80.69 for ANN, SVM, and DBN, respectively.

This article involves the use of value stream mapping (VSM) and simulation to improve the efficiency of the batch production system commonly found in small and medium enterprises. VSM is a tool for portraying the system to identify areas for improvement. The simulation model imitates a real production system so that improvement alternatives can be evaluated without disrupting actual production. The simulation can also optimise the levels of resources required for the bottleneck operations. The production process at a roasted and ground coffee plant that had serious capacity shortage problem is used as a case study. Three improvement alternatives are evaluated: (1) increasing the workforce at the bottleneck operations; (2) using a new workforce management plan formulated from analysing the VSM and simulation and (3) investing in a machine to replace the bottleneck manual operations. Analysis results show the capability of these tools to formulate effective solutions for the case study system.

Physical distribution is one of the key functions in logistics systems, involving the flow of products from manufacturing plants or distribution centers through the transportation network to consumers. It is a very costly function, especially for the distribution industries. While maintaining the desired customer service levels, an effort is made in this paper to improve distribution strategies and reduce the distribution cost for the multi-product, multi-depot periodic distribution problem. In industry practice, depots typically operate independently and solely within their own territories. However, it may be beneficial to allow those depots to operate interdependently, particularly when the product supplies are limited at some depots. In such cases, the distributors may satisfy customers' requests by delivering products from other depots that hold more supplies. In particular, the impact of interdependent operations among depots, which has not previously been addressed in the context of industrial applications, is investigated in this research. A mixed-integer linear programming model is formulated to represent the multi-product, multi-depot periodic distribution problem. Three tabu search heuristics with different long-term memory applications are developed to solve the problem. The performance of the heuristics is evaluated by comparing the solutions obtained with the optimal solutions or lower bounds from the regular branch-and-bound method as well as a fast technique to find a lower bound that is developed in this research. The heuristics provide optimal/good quality solutions in a much shorter time. A randomized complete block design is applied to test the performance of the heuristics on various problem structures. The experimental results show that the tabu search heuristic that incorporates the use of a long-term memory in the diversification process outperforms the other heuristics. The heuristic is further applied to investigate the impact of interdependent operations among depots. The results reveal that interdependent operations among depots provide significant savings in costs over independent operations among depots, especially for large-size problems.