Análisis predictivo de la demanda de bebidas para la estimación del inventario dormido en un establecimiento hotelero.

Abstract

This research analyzes the temporal behavior of demand for slow-moving beverages in a hotel establishment in the city of Guayaquil, with the main objective of projecting the level of dormant inventory for the year 2026. Storing products with no movement represents a critical problem that results in immobilized capital, a high risk of obsolescence, and higher storage costs, which directly affects the operational efficiency of the inventory department. Using a quantitative, descriptive, and predictive approach, historical data from the years 2024 and 2025 were used. After a process of cleaning and consolidating all monthly requisitions, time series were constructed to obtain the dormant inventory, which is operationally defined as any product with no movement during a consecutive quarter. The turnover analysis performed by categories, such as beers, wines, and spirits, allowed for the identification of stationery and variability patterns. Within the forecasting phase, a predictive model, SARIMA, was implemented and validated using mean error metrics (MAE) and significance analysis. The results show that low product turnover exhibits seasonal variations linked to fluctuations in demand at the facility, while maintaining a stable average. Projections for 2026 indicate the persistence of immobilized stock under current policies, emphasizing the need to integrate predictive models into planning. The study concludes that time series analysis is a fundamental resource for mitigating financial risks in the context of fluctuating demand.