Time series regression modeling and prediction of book borrowing volume with time-space-interest psychological factors

Abstract

This study analyzes the borrowing volume of 5,272 male and 15,935 female readers from Nanjing Normal University during 2022-2024. Conducting a classification statistic across 12 months, 22 library locations, and 23 major book categories, it extracts common factors in time, space, and interest dimensions via factor analysis. Using the standardized scores of these factors as independent variables and the logarithmically transformed borrowing volume as the dependent variable, regression analysis is performed to assess the influence of each dimension, emphasizing the significance of time. A time series model is applied to explore the monthly average change patterns of borrowing volume from 2022 to 2024 and predict the per capita borrowing volume for each month in 2025. Results indicate distinct gender differences. Female temporal factors show significant seasonal clustering, while male temporal factors remain consistent throughout the year. Females' spatial patterns are more regionally fixed, whereas males exhibit multi-dimensional and specialized spatial selections. Female reading interests center on humanities, while males' interests are distributed across theory, history, and application. The predicted per capita borrowing volumes in 2025 for males range from 2.84 (August) to 4.13 (March), and for females, from 2.49 (August) to 3.48 (July), with upper and lower control limits provided for each month, offering insights into future borrowing trends.