基于增强回归树的水藻预测分析

Forecast and Analysis of Algae Based on Boosted Regression Tree

河流水质的预测分析在保护河流水源和维护河流生态有着重要意义。由于基于多元线性回归没有处理数据缺失值的能力和决策树模型无法有效处理水质多变量的问题等原因,故两者均达不到有效预测水质影响因素的目标。本文采用的增强回归树模型能够处理缺失值和避免过度拟合问题,可以有效地对水质的藻类进行预测分析并得出综合影响测试河流中综合影响7种藻类繁殖的主要因素。实验分析结果表明,采用的增强回归树模型优于多元线性回归模型。

Forecast and analysis of water quality of rivers play an important role in the protection of water sources and the maintenance of ecology. Because the multivariate linear regression can not deal with the missing values and the model of decision trees can not deal with multiple variables of water data, the goal of forecasting the influencing factors of water quality can not be achieved effectively. In this paper, the boosted regression tree（BRT） model is used to solve the problem of the missing values and avoid over fitting, which a- vailably forecasts the main factors influencing the reproduction of seven algae of the tested rivers. Experiments indicate that BRT per- forms better than multivariate linear regression.