第二周<线性回归>

可行性分析####

略

sklearn.linear_model.linear_regression()

一些参数

• fit_intercept 布尔型参数，表示是否计算该模型的截距
• normalize 布尔型参数，若为TRUE，则X在回归前进行归一化，默认False

可以查看系数###

linear.coef_

linear.intercept_

# -*- coding: utf-8 -*-
"""
Created on Sat May 27 12:04:03 2017
@author: sfzyk
"""
import numpy as np
import matplotlib.pyplot as plt
import os
from sklearn.linear_model import LinearRegression
os.chdir(r"d:\mechine_learning\mooc_data")
f=open("prices.txt",'r')
lines=f.readlines()
data_x=[]
data_y=[]
for line in lines:
    #print(line)
    items=line.strip()
    strs=items.split(',')
    data_x.append(int(strs[0]))
    data_y.append(int(strs[1]))
leng=len(data_x)
data_x=np.array(data_x).reshape((leng,1))
#同是一维的但是data_x 和data_y的格式要求不一样
data_y=np.array(data_y)
minx=min(data_x)
maxx=max(data_x)
x=np.linspace(minx,maxx,100)
linear=LinearRegression()
linear.fit(data_x,data_y)
plt.scatter(data_x,data_y,'r')
plt.plot(x,linear.predict(x.reshape(-1,1)),'-b')

加入高次项特征###

# -*- coding: utf-8 -*-
"""
Created on Sat May 27 12:59:12 2017
@author: sfzyk
"""
from sklearn.preprocessing import PolynomialFeatures
import matplotlib.pyplot as plt
import os
import numpy as np
from sklearn.linear_model import LinearRegression
os.chdir("d:\mechine_learning\mooc_data")
f=open("prices.txt",'r')
lines=f.readlines()
data_x=[]
data_y=[]
for line in lines:
    #print(line)
    items=line.strip()
    strs=items.split(',')
    data_x.append(int(strs[0]))
    data_y.append(int(strs[1]))
leng=len(data_x)
data_x=np.array(data_x).reshape((leng,1))
minx=min(data_x)
maxx=max(data_x)
x=np.linspace(minx,maxx,100)
poly_reg=PolynomialFeatures(degree=2)
x_poly=poly_reg.fit_transform(data_x)
linear=LinearRegression()
linear.fit(x_poly,data_y)
plt.scatter(data_x,data_y,color='red')
plt.plot(x,linear.predict(poly_reg.fit_transform(x.reshape((-1,1)))),'-b')