论文部分内容阅读
城市用水量受到多重因素的影响,且各因素之间存在相关性。将逐步回归技术引入偏最小二乘(PLS)用水量预测模型影响因子的筛选过程,可对PLS回归建模过程进行改进,在保证拟合精度的条件下,有效解决了自变量间的多重相关性问题;同时实现测定指标的降维,达到了简化、精炼模型的目的。将所提理论和方法应用于某城市用水量预测中,运用R软件进行求解,并将耦合逐步回归的PLS模型与单一的PLS回归模型进行比较分析。结果表明,模型的拟合和预报精度较好。
Urban water consumption is affected by multiple factors, and there is a correlation between the factors. The step-by-step regression technique was introduced to select the influence factors of partial least squares (PLS) water consumption prediction model, which improved the PLS regression modeling process and effectively solved the multiple correlations among the independent variables Sexual problems; at the same time to achieve measurement dimension reduction, to achieve the purpose of simplifying and refining the model. The proposed theory and method are applied to the prediction of water consumption in a city. The software R is used to solve the problem. The PLS model with stepwise regression is compared with a single PLS regression model. The results show that the model fitting and forecasting accuracy is better.