基于Lasso和支持向量机的组合预测及其应用

Combination Forecast and its Application Based on Lasso and Support Vector Machine

针对Lasso方法与支持向量机两者的联系与各自的优势,给出了基于Lasso与支持向量机的串联型、并联型和嵌入型三种组合预测,并将它们运用到我国粮食价格预测中.实证结果表明,与单一预测方法的预测结果相比,基于Lasso方法与支持向量机的串联型组合预测和嵌入型组合预测具有更高的预测精度.

In view of the connection between Lasso and support vector machine and their respective advantages,we gave three kinds of combination forecast model,which includes series combination forecast,parallel combination forecast and embedded combination forecast.Then we used them in China’s grain price forecast.The empirical results show that compared with the prediction results of single prediction method both series combination forecast and embedded combination forecast based on Lasso and Support Vector Machine have higher prediction accuracy.