SUPPORTING THE INVENTORY MANAGEMENT IN THE MANUFACTURING COMPANY BY CHATGPT

Background: The decision-making process in the operational context of enterprises is an integral aspect of how they function, and in this area, precise demand forecasting plays a key role. The use of accurate forecasting models not only meets customer expectations but also enables efficient resource allocation and operational cost optimization. In the long term, such actions contribute to increasing the organization's competitiveness in the market. In recent years, there has been a growing trend in the use of advanced analytical technologies, including machine learning, for demand forecasting purposes. This scientific paper focuses on a comparative analysis of demand forecasting effectiveness using the generative language model GPT in relation to the auto ARIMA algorithm.

Methods: A case study analysis for a selected manufacturing organization was conducted based on twelve diversified operational references, for which the supply chain mechanisms are heterogeneous. In the research process, a classification into four reference groups was established, based on the time required to complete the ordering process. Forecast generation was carried out using the auto.arima() algorithm in the R programming environment, as well as through the ChatGPT language model versions 3-5. The forecast results were subjected to comparative analysis, in which weighting was applied for different forecast accuracy indicators, including the Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and the number of precisely predicted daily forecasts.

Results: The study showed that ChatGPT is more reliable in forecasting compared to ARIMA. However, integrating ChatGPT into the existing systems in the company can be problematic, mainly due to limitations in data operations. Despite this, ChatGPT has the potential to improve the accuracy of inventory management plans both in the short and long term.

Conclusions: The comparative analysis of the effectiveness of forecasting models, including ChatGPT and ARIMA, showed that the ChatGPT algorithm achieves higher levels of forecasting accuracy. This is observed despite increased computational complexity and challenges associated with processing large data sets.