Optimizing business strategies for energy through machine learning involves using predictive analytics for accurate energy demand and price forecasting, enhancing operational efficiency through resource optimization and predictive maintenance, and optimizing renewable energy integration into the energy grid. This approach maximizes production, reduces costs, and ensures stability in energy supply. The novelty of integrating deep reinforcement learning (DRL) in energy management lies in its ability to adapt and optimize operational strategies in real-time, autonomously leveraging advanced machine learning techniques to handle dynamic and complex energy environments. The study’s outcomes demonstrate the effectiveness of DRL in optimizing energy management strategies. Statistical validity tests revealed shallow error values [MAE: 1.056 × 10(− 13) and RMSE: 1.253 × 10(− 13)], indicating strong predictive accuracy and model robustness. Sensitivity analysis showed that heating and cooling energy consumption variations significantly impact total energy consumption, with predicted changes ranging from 734.66 to 835.46 units. Monte Carlo simulations revealed a mean total energy consumption of 850 units with a standard deviation of 50 units, underscoring the model’s robustness under various stochastic scenarios. Another significant result of the economic impact analysis was the comparison of different operational strategies. The analysis indicated that scenario 1 (high operational costs) and scenario 2 (lower operational costs) both resulted in profits of $70,000, despite differences in operational costs and revenues. However, scenario 3 (optimized strategy) demonstrated superior financial performance with a profit of $78,500. This highlights the importance of strategic operational improvements and suggests that efficiency optimization can significantly enhance profitability. In addition, the DRL-enhanced strategies showed a marked improvement in forecasting and managing demand fluctuations, leading to better resource allocation and reduced energy wastage. Integrating DRL improves operational efficiency and supports long-term financial viability, positioning energy systems for a more sustainable future.

Data commentary is an important text type in research articles; however, its discourse model is often challenging to access because it is embedded in the upper genres such as textbook, weather forecast, and journal article. This study aims to establish a discourse model of data commentary, with a focus on academic research papers in Economics and Business administration journals. To accomplish this, this study employs Move analysis and SF-MDA(Systemic Functional-Multimodal Discourse Analysis) to investigate the moves of data commentaries and the metafunctional meanings of each step. The results indicate that the data commentary discourse model consists of three moves: (1) summarizing the topic and methodology, (2) representing figure and numbers, and (3) analyzing and commenting on results. Additionally, 22 steps are identified for each move that creates metafunctional meaning: ideational, interpersonal, and textual.