Book borrowing behavior analysis and prediction based on time series analysis and LightGBM

Through detailed time series analysis of historical book borrowing records, this paper designs a specific time window to adapt to the timing of time series data. Since XGBoost has good explanatory and debugging capabilities, it uses XGBoost for preliminary feature engineering and pre-processing. We input its optimized features into LightGBM for final modeling, and use its characteristics of high speed and low memory consumption to deal with nonlinear relations and high-dimensional feature space problems in borrowing data. This method skillfully combines LightGBM and XGBoost, and integrates the essence of time series analysis. It can construct a strong prediction model that not only considers the complexity of time series characteristics, but also captures the nonlinear relationship effectively, and brings more accurate and comprehensive insight to the analysis of book borrowing behavior.