读者驱动资源建设中预测算法的研究及应用

PREDICTION ALGORITHM AND ITS APPLICATION IN READER-DRIVEN RESOURCE CONSTRUCTION

针对读者驱动资源建设策略中的合理预测这一关键问题进行研究。分析已有数据特征,建立不同周期模式下的资源预测机制。分别建立以月为周期的BP神经网络预测算法,以学期为周期的GM(1,1)灰色预测算法,对未来一段时间内各类图书借阅人数进行预测。使用相对误差分析法检验法分别检验了两种预测机制的合理性,最后进行横向对比并给出合理化建议。该方法可应用于对更多资源的综合预测及评价。

This paper studied the key problem of reasonable prediction in reader-driven resource construction strategy. We analyzed the characteristics of existing data and established resource prediction mechanism under different cycle modes. We established BP neural network prediction algorithm with monthly cycle and GM(1,1) grey prediction algorithm with semester cycle to predict the number of books borrowed in the future. We used the relative error analysis method to test the rationality of the two prediction mechanisms, and compared them horizontally and gave reasonable suggestions. This method can be applied to the comprehensive prediction and evaluation of more resources.