Validation 学习曲线
我在第6周,在Coursera的机器学习课程中做练习5。你能帮我解决我的两个问题吗 1/我得到了错误的训练曲线(蓝色)和验证曲线(橙色),如下所示。(我对θ值使用正规方程。我如何修正它Validation 学习曲线,validation,machine-learning,Validation,Machine Learning,我在第6周,在Coursera的机器学习课程中做练习5。你能帮我解决我的两个问题吗 1/我得到了错误的训练曲线(蓝色)和验证曲线(橙色),如下所示。(我对θ值使用正规方程。我如何修正它 J_train = numpy.zeros((12, 1)) J_cv = numpy.zeros((12, 1)) NumberOfExamples = numpy.zeros((12, 1)) for i in range (12): Xtrain = X[0:i+1] Ytrai
J_train = numpy.zeros((12, 1))
J_cv = numpy.zeros((12, 1))
NumberOfExamples = numpy.zeros((12, 1))
for i in range (12):
Xtrain = X[0:i+1]
Ytrain = y[0:i+1]
theta = inv(Xtrain.T * Xtrain ) * Xtrain.T * Ytrain
J_train[i] = (1/(2*m)) * numpy.sum(numpy.power((Xtrain * theta - Ytrain) , 2))
J_cv[i] = (1/(2*m)) * numpy.sum(numpy.power((Xval * theta - yval) , 2))
NumberOfExamples[i] = i + 1
plt.plot(NumberOfExamples, J_train, NumberOfExamples, J_cv)
但应该是这样的:
2/在这种情况下,为了计算训练示例的成本和验证集的成本,我们为什么不像往常一样在X(值为1的列)和θ(s)中添加“偏差”
多谢各位