[目的]用ARIMA季节乘积模(p,d,q)(P,D,Q)s对季节性时间序列资料建模并预测。[方法]用Box-Ljung统计量评价ARIMA模型的拟和度,用平均预测相对误差作为预测效果的评价指标。[结果]对所分析的季节性时间序列建立了乘积ARIMA(0,1,1)(0,1,2)s模型,平均预测相对误差为6.50%。[结论]用所建立模型对入院人次进行预测,结果表明ARIMA是一种预测精度较高的预测模型。 [Objective] The ARIMA seasonal product model (p, d, q) (P, D, Q) s is used to model and predict seasonal time series data. [Method] The Box-Ljung statistic was used to evaluate the fitting degree of ARIMA model. The relative error of average prediction was taken as the evaluation index of prediction effect. [Result] The product ARIMA (0,1,1) (0,1,2) s model was established for the analyzed seasonal time series. The average relative prediction error was 6.50%. [Conclusion] Using the established model to predict hospital admission, the results show that ARIMA is a prediction model with high prediction accuracy.