Electric vehicles are attracting attention from the perspective of environmental protection. However, their short cruising range and insufficient charging infrastructure have hindered their widespread use. The home delivery industry is less affected by these characteristics, and demonstration experiments are already underway to replace delivery trucks with electric vehicles. Various studies have been conducted on electric vehicle routing problems. In most of these studies, delivery routes have been considered focusing on the travel distances of the electric vehicles. Recently, the problem of electricity supply has been changing due to the spread of solar power generation. Electricity market prices and carbon dioxide emissions during electricity generation now vary dramatically according to the time of day . The present paper considers the problem of minimizing costs and carbon dioxide emissions in electric vehcles used for deliveries, rather than the problem of the travel distance. The present paper proposes a mathematical model that extends the model of Schiffer and Walther〔1〕to calculate charging costs and carbon dioxide emissions even in situations where electricity prices and carbon dioxide emissions depend on the time of day. It is possible to show appropriate routes by using the proposed mathematical model. In addition, the tradeoff among charging costs, carbon dioxide emissions, and travel distance is also discussed. As an application example of the model, the characteristics of the charging stations used and the reductions in charging costs and carbon dioxide emissions from electric vehicles in numerical experiments assuming real regions are also shown.

In recent years, online shopping sites have implemented various business measures to improve profitability, including coupon issuance and point redemption. To optimize these measures and maximize profits, managers must set coupon discount amounts and point redemption amounts. One approach to solving this problem is to use machine learning to estimate a function with the inputs of business measures and the output of outcome variables. However, the relationship between the input and the output is not known in advance, and there is no training data for estimating the function before a measure is implemented. However, since the purpose of a business is to make a profit, it is often difficult to conduct large-scale experiments on real businesses where the only purpose is to acquire such data. On the other hand, Bayesian optimization is attracting attention as a method for performing sequential optimization of input while sequentially adding training data to an unknown function. Bayesian optimization estimates the posterior distribution of the output from the training data and uses an evaluation index called the acquisition function to estimate the next data point that will optimize the input. However, ordinary Bayesian optimization may not produce appropriate results when applied to practical business because it does not consider the characteristics of business effects, such as differences in variance depending on the input. Therefore, this study proposes a new acquisition function for Heteroskedastic Gaussian Process (hetGP), a function estimation method with different noise variances, that can consider the unique circumstances of business measures. This paper uses artificial data to demonstrate that the proposed method can effectively optimize business policies, even for functional data with input-dependent error variance that has not been handled by Bayesian optimization before. This method can enable regular optimization of business measures.

If network visualization can be achieved from the conversation data accumulated on business chat apps, which has recently been utilized, the actual state of communication within workplace teams for business purposes can be clarified, and it has the potential to be useful for various measures and decision making. In the network analysis that visualizes communication among employees, each employee is represented by a node and the communication statuses are represented by links between nodes. Although the employees represented by each node can be uniquely defined for a target company, the relationship between nodes can be variously defined from the different viewpoints and analysis purposes. That is, the meaning of the links changes depending on the conversation conditions (communication conditions) used to define edges, such as the content of statements, topics, and groups to which the employees belong. It would be beneficial from the viewpoint of organizational management if a method for analyzing differences and similarities in network structures due to such differences in conditions was provided. This study proposes a network analysis method for communication data on business chat apps, which quantitatively analyzes and visualizes changes in employee network structure due to differences in communication conditions. The proposed method makes it possible to visualize changes in network structure associated with different communication conditions and communication conditions with structures that differ significantly from the overall communication structure in a form that is easy to interpret and can be used for various measures and decision making. This study demonstrates the effectiveness of this method for analyzing relationships among employees by using actual conversation history data.

Gas burning that occurs during the resin molding process is fundamentally due to the improper venting of gas. However, finding a solution is extremely difficult as the phenomenon arises from a complex interplay of numerous factors. In the maintenance of molding machines and molds, carrying out regular inspections, replacing parts, cleaning, and maintaining the machine condition are practical and effective measures. During quality control, it is important to perform maintenance at the appropriate timing, but the timing is often determined based on experience and intuition, meaning that there is a lack of clear standards. This paper proposes a method using a deep learning classification model to clarify the timing at which maintenance should be performed based on the perspectives of quality and efficiency. In the proposed method, first, the cosine similarity between images of an initially molded product, which serves as a standard, and images of each subsequent product is calculated. Then, transfer learning of the deep learning model is performed in which classification is performed based on similarity. Finally, fine-tuning is performed. Emphasis is placed on the utilization of a loss function that directly controls the feature values during the fine-tuning.