overfitting problem &underfitting problem, ti's just not even fitting the training data very well.
Overfitting-----------
a technique called regularization that will allow us to amelioratr or to reduce this overfitting problem and get these learning algorithms to maybe work much better.this algorithm has high variance.
使用高阶函数能够好的拟合数据,但是在测试数据上可能会效果不好,因为产生了过拟合,即泛化能力不好。
产生过拟合的一种情况是:训练集少,但是特征变量特别多(因为每一个特征变量,都能对预测的y值产生一点影响)。所以第一种可以通过throwing away some of the features. 但是有时间这些特征是非常有用的,这个时候可以通过第二种引入正则化(regularization),Here,we're going to keep all the features,but we are going to reduce the magnitude of the values of the parameters theta j.