Programming Assignment的心得（一）

Linear Regression and Logistic Regression

主要代码

class linear_regression:

	def __init__(self):
		self.mse_func = lambda x,y: 1/x.shape[0] * np.sum((x-y)**2)
		pass

	def train(self,x_train, y_train):
		"""Receive the input training data, then learn the model.
		Inputs:
		x_train: np.array, shape (num_samples, num_features)
		y_train: np.array, shape (num_samples, )

		Outputs：
		None
		"""

		self.learning_rate = 0.0001
		self.w = np.random.rand(10)
		iteration = 10000

		for i in range(iteration):
			pred = self.predict(x_train)

			"""
			Please Fill Your Code Here.
			"""
            gradient = 2 * 1/x_train.shape[0] * np.dot(x_train.T, (pred - y_train))
			self.w = self.w - self.learning_rate * gradient
			
			if i % 100 == 0:
				print("Iteration {}/{}, MSE loss {:.4f}".format(i+1, iteration, self.mse_func(pred, y_train)))

		return

	def predict(self, x_test):
		"""Do prediction via the learned model.
		Inputs:
		x_test: np.array, shape (num_samples, num_features)

		Outputs:
		pred: np.array, shape (num_samples, )
		"""
		
		"""
		Please Fill Your Code Here.
		"""
        pred = np.dot(x_test, self.w)
        
		return pred

缩进问题

问题：spyder打开作业的py文件后，自己写的代码缩进无法一致

解决：

• 进入tools-preference
• 查看editor的display，选择第二项show blank spaces，发现源代码的空格显示为$\to$，而自己的代码空格处显示为$\cdots$
• 选择editor的advanced settings 将indentation characters从4 spaces改为tabulations，同时勾选下面的tab always indent

矩阵形式写梯度

Theory：

for the ordinary least square problem:
$$\begin{array}{rl}J(\theta )=\frac{1}{2}\sum _{i=1}^m{\left(h\left(x^{(i)}\right)-y^{(i)}\right)}^2=& \frac{1}{2}\sum _{i=1}^m(\end{array}$$