要建立一个有效的支持向量回归(SVR)模型,支持向量回归的3个参数C,γ,ε必须预先设定。提出一种新型的遗传算法——智能遗传算法(IGA)对支持向量回归进行参数调节,以达到寻找最优参数的目的,然后和支持向量回归结合得到一种新的IGASVR模型,并应用于城市人口预测。最后,将提出的方法与标准SVR模型和BP神经网络模型进行比较,所得结果表明,该模型训练速度快,并且有较高预测精度,是一种有效的人口预测方法。
To build an effective SVR model,SVR's parameters must be set carefully.This study proposes a novel approach, known as IGASVR,which searches for SVR's optimal parameters using intelligent genetic algorithms,and then adopts the optimal parameters to construct the SVR models.Finally we apply IGASVR to forecast population.The experimental results demonstrates that IGASVR are better than standard SVR and BP neural-network.IGASVR model is an effective approach which has faster speed of training and higher precision.