线性回归简介
线性回归是机器学习中最简单和最常见的回归问题之一。它是一种用于预测连续数值输出的统计模型,它假设自变量(输入特征)与因变量(输出)之间存在线性关系。
在线性回归中,我们假设因变量与自变量之间的关系可以表示为一个直线方程:y=wx+b,其中w、b为模型的相关参数,x为输入特征,y为预测值。
训练线性回归模型的步骤包括以下几个关键步骤:
-
数据收集:收集包含自变量和因变量的样本数据。这些数据可以是实际观测值或者从其他来源获取的数据。
-
数据预处理:对数据进行预处理,包括数据清洗、缺失值处理、特征工程等步骤,以便为模型提供可用的数据。
-
特征选择:选择与因变量相关的自变量作为模型输入特征。通过研究特征之间的相关性和各自与因变量的关系,来决定哪些特征对预测较为重要。
-
模型训练:使用训练数据拟合线性模型,并找到最佳的斜率和截距。通常采用最小二乘法或者梯度下降等优化算法来最小化模型的残差。
-
模型评估:使用评估指标(例如均方误差、决定系数等)来评估模型的性能。一般来说,较低的均方误差和较高的决定系数表示模型预测效果较好。
-
模型应用:使用训练好的模型进行预测。给定新的自变量,模型可以通过线性方程预测相应的因变量。
代价函数
公式如下:
代码:
import numpy as np
import matplotlib.pyplot as plt
from lab_utils_uni import plt_intuition, plt_stationary, plt_update_onclick, soup_bowl
plt.style.use('./deeplearning.mplstyle')
def compute_cost(x,y,w,b):
"""
:param x: data,m examples
:param y: target values
:param w: model parameters
:param b: model parameters
:return: total_cost: The cost of using w,b as the paramters to fit the data points in x and y
"""
m = x.shape[0]
cost_sum = 0
for i in range(m):
f_wb = w * x[i] + b
cost = (f_wb - y[i]) ** 2 #两个*表示平方
cost_sum = cost_sum + cost
total_cost = (1/(2*m))*cost_sum
return total_cost
x_train = np.array([1.0, 1.7, 2.0, 2.5, 3.0, 3.2])
y_train = np.array([250, 300, 480, 430, 630, 730,])
plt.close('all')
fig, ax, dyn_items = plt_stationary(x_train, y_train)
updater = plt_update_onclick(fig, ax, x_train, y_train, dyn_items)
soup_bowl()
运行效果:
可以看到当cost(w,b)越靠近中间圆圈,预测值越接近于真实值
梯度下降代码
具体运算公式如下图
代码:
import math, copy
import numpy as np
import matplotlib.pyplot as plt
from lab_utils_uni import plt_house_x, plt_contour_wgrad, plt_divergence, plt_gradients
# Load our data set 数据集
x_train = np.array([1.0, 2.0]) #features
y_train = np.array([300.0, 500.0]) #target value
# Function to calculate the cost 代价函数
def compute_cost(x, y, w, b):
m = x.shape[0]
cost = 0
for i in range(m):
f_wb = w * x[i] + b
cost = cost + (f_wb - y[i]) ** 2
total_cost = 1 / (2 * m) * cost
return total_cost
# Function to calculate the gradient
def compute_gradient(x,y,w,b):
"""
:param x: Data, m examples
:param y: target values
:param w: model parameter
:param b: model parameter
:returns
dj_dw: The gradient of the cost w.r.t The parameters w
dj_db: The gradient of the cost w.r.t The parameters b
"""
m = x.shape[0]
dj_dw = 0
dj_db = 0
for i in range(m):
f_wb = w * x[i] + b
dj_dw_i = (f_wb - y[i]) * x[i]
dj_db_i = (f_wb - y[i])
dj_dw = dj_dw + dj_dw_i
dj_db = dj_db + dj_db_i
dj_dw = dj_dw / m
dj_db = dj_db / m
return dj_dw, dj_db
plt_gradients(x_train,y_train, compute_cost, compute_gradient)
plt.show()
def gradient_descent(x, y, w_in, b_in, alpha, num_iters, cost_function, gradient_function):
"""
:param x: Data, m examples
:param y: target values
:param w_in,b_in: initial values of model parameters
:param alpha: Learning rate 学习率
:param num_iters: number of iterations to run gradient descent 迭代次数
:param cost_function: function to call to produce cost
:param gradient_function: function to call to produce gradient
:Returns:
w (scalar): Updated value of parameter after running gradient descent
b (scalar): Updated value of parameter after running gradient descent
J_history (List): History of cost values
p_history (list): History of parameters [w,b]
"""
w = copy.deepcopy(w_in) #深复制避免赋值后原变量被修改
J_history = []
p_history = []
b = b_in
w = w_in
for i in range(num_iters):
#计算梯度
dj_dw, dj_db = gradient_function(x,y,w,b)
#更新参数
b = b - alpha * dj_db
w = w - alpha * dj_dw
if i<100000:
J_history.append(cost_function(x,y,w,b))
p_history.append([w,b])
# Print cost every at intervals 10 times or as many iterations if < 10
if i % math.ceil(num_iters / 10) == 0:
print(f"Iteration {i:4}: Cost {J_history[-1]:0.2e} ",
f"dj_dw: {dj_dw: 0.3e}, dj_db: {dj_db: 0.3e} ",
f"w: {w: 0.3e}, b:{b: 0.5e}")
return w,b,J_history,p_history
# initialize parameters
w_init = 0
b_init = 0
# some gradient descent settings
iterations = 10000
tmp_alpha = 1.0e-2 #学习率设置不宜过大也不宜过小
# run gradient descent
w_final, b_final, J_hist, p_hist = gradient_descent(x_train ,y_train, w_init, b_init, tmp_alpha,
iterations, compute_cost, compute_gradient)
print(f"(w,b) found by gradient descent: ({w_final:8.4f},{b_final:8.4f})")
#plot cost versus iteration
fig, (ax1, ax2) = plt.subplots(1, 2, constrained_layout=True, figsize=(12,4))
ax1.plot(J_hist[:100])
ax2.plot(1000 + np.arange(len(J_hist[1000:])), J_hist[1000:])
ax1.set_title("Cost vs. iteration(start)"); ax2.set_title("Cost vs. iteration (end)")
ax1.set_ylabel('Cost') ; ax2.set_ylabel('Cost')
ax1.set_xlabel('iteration step') ; ax2.set_xlabel('iteration step')
plt.show()