Intelligent Mobile Budget Forecasting Using LSTM Networks: A Dynamic Decision-Support Framework for Emerging Markets

Budget forecasting is essential for organizational resource allocation and risk-aware decision-making, yet conventional models often struggle with nonlinear patterns and rapidly changing operating conditions. This study proposes an intelligent mobile decision-support framework using long short-term memory (LSTM) neural networks for dynamic budget forecasting. The framework integrates historical financial data, macroeconomic indicators, and operational metrics through a dual-path LSTM architecture and an adaptive sliding-window mechanism. The model is evaluated using manufacturing and retail datasets and benchmarked against SARIMA and XGBoost. Results demonstrate improved forecasting accuracy and resilience under changing conditions, achieving a 19.2% higher Forecast Skill Score in manufacturing and a 15.7% improvement in retail scenarios. The framework supports mobile-accessible managerial decision-making by transforming dynamically updated forecasts into actionable information for resource allocation and operational risk management. Its scalable architecture is particularly relevant to digitally evolving emerging markets, including Indonesia, where organizations increasingly require accessible and adaptive decision-support technologies. The study contributes an AI-enabled approach that integrates dynamic forecasting with mobile decision support, organizational resilience, and sustainable resource utilization.