Research on Complex System Flow Prediction Modeling Based on Linear Methods

Aiming at the practical demand of passenger flow arrangement and resource scheduling for amusement park operation, this paper takes a multi-factor-driven complex passenger flow system as the research object and constructs a prediction model based on vector space theory and matrix fitting of linear algebra. Taking multiple linear regression as the core modeling framework, this study quantifies the influence of temperature, weather, holiday type, and calendar factors on daily visitor volume through matrix operation and gradient descent optimization. The dataset covers five consecutive years (2021–2025) of daily operation records from a medium-sized domestic amusement park, which is divided into training and test sets for model training and verification. Experimental results show that the proposed linear model achieves an average absolute error (MAE) lower than 6% under normal weather and holiday conditions, with the coefficient of determination R2=0.942, which outperforms the traditional empirical forecasting method whose MAE reaches 22.7%. The model has low computational cost and strong interpretability and can provide quantitative data support for daily staffing and material reserve planning of scenic spots. However, the model produces obvious prediction deviations exceeding 25% under extreme weather or temporary large-scale park activities with strong nonlinear characteristics, which points out the direction for subsequent model optimization.