Year: 2025
Type: article
Abstract: This paper is a predictive analytics system in the planning of art festivals through the development of complex statistical and machine learning models. The increasing complexity of the festival management due to the variability of attendance, fluctuating budgets, and other external variables like weather and tourism demand data-driven decision support systems. The study draws upon various sources of data including statistics of events in the past, ticket sales, social media activity, and weather forecasts to model key performance measures of attendance, logistical demand, and financial efficiency. They use three predictive layers trend analysis (ARIMA and Prophet as time-series forecasting models), structured data-based insights (machine learning algorithms, such as Random Forest and XGBoost), and temporal-sequential pattern recognition (deep learning architectures, such as LSTM and Transformer predictors). These results prove that the hybrid ensemble models are better than the single model ones, with an accuracy of up to 14 percent higher on attendance prediction and 11 percent better on resource allocation optimization. Results indicate that early ticketing, real-time weather information, and online sentiment dynamic have a predictive effect on the attendance of the festival and cost-related stability. This structure offers the administrative bodies of festivals and culture a sound data-wise instrument of strategic planning, marketing and planning of operations in real-time.
Field: Decision Sciences
Domain: Social Sciences
Open Access status: diamond
OpenAlex ID: W7117407202