SPSS在医院统计预测中的应用

Application of SPSS in Hospital Statistical Prediction

目的 探讨SPSS 8.0统计软件包中回归、指数平滑及ARIMA等时间序列分析模块的建模及诊断方法。方法 根据某医院 1995年 1月— 2 0 0 1年 11月的门诊人次历史资料 ,建立对数模型、指数平滑模型和ARIMA乘积模型 ,并对三者的预测结果进行比较分析。结果 对数模型、指数平滑模型和ARIMA乘积模型的预测平均相对误差分别为 14 .34% ,8.14 %和 4 .89%。结论 ARIMA乘积模型适于对有趋势性和周期性的门诊量数据进行预测。SPSS 8.0统计软件包时间序列分析模块操作方便 。

Objective To sum up the modulars of statistical prediction in SPSS 8.0 which are regression, exponential smoothing and ARIMA and analyze the methods to model and diagnose these models. Methods We established logarithmic model, exponential smoothing model and model of seasonal multiple ARIMA and compared the prediction effects of them. Results The average relative error of logarithmic model, exponential smoothing model and model of seasonal multiple ARIMA was 14.34%, 8.14% and 4.89%, respectively. Conclusion The model of multiple seasonal ARIMA can be well used to predict the time series of outpatient amount which has the trend and periodicity. Time series analysis modulars can be efficiently applied to hospital statistical prediction.