我的数据似乎比线性回归更适合多项式回归,但Azure Machine Learning Studio没有多项式回归的原生支持 . Is there a way to transform data and produce a web service capable of predicting values that fit this type of data?

我找到了一个很好的article here,讨论了如何训练模型以拟合多项式回归,即使Azure ML确定的系数为0.973346 . 但是,当我创建"predictive experiment"和Web服务来预测新值,并要求服务预测新值时,准确性是可怕的,就像在球场中一样 .

How can I train a predictive model to data that fits a polynomial regression using Azure Machine Learning Studio?