城市日用水量预测模型及其应用

Prediction Model of Urban Daily Water Consumption and Its Application

提出成分数据赋权指数平滑模型,将该模型应用到城市日用水量预测中,用平均绝对百分比误差来表征其预测精度,结果显示,其平均绝对百分比误差MAPE为2.2%,表明该模型的预测精度较高。并将其与直接应用赋权指数平滑模型进行比较,结果显示,成分数据方法的引用使预测精度提高了0.18%,分析可知成分数据的转化与数据的重新排列使数据变化趋势趋于平稳,有利于提高预测精度。

A weighted exponential smoothing model based on the compositional data was introdueed for predicting urban daily water consumption, and the mean absolute percent error （MAPE） characterized forecasting accuracy. The results show that the mean absolute percent error （MAPE） is 2.2%, indicating the model has preferable precision. The weighted exponential smoothing model based on the compositional data was compared with weighted exponential smoothing model. As a result, the prediction accuracy is further increased by 0.18 %. Analysis shows that compositional data and proper data processing make the data variation trend to he steady, and is beneficial to the improvement of prediction accuracy.