《用Python玩转数据》项目—线性回归分析入门之波士顿房价预测（一）

from sklearn import datasets
from sklearn import linear_model
from sklearn.cross_validation import train_test_split
from sklearn import metrics
import os
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np


'''----------load 数据集-----------'''
dataset = datasets.load_boston()

# x 训练特征：['CRIM', 'ZN', 'INDUS', 'CHAS', 'NOX', 'RM', 'AGE', 'DIS',
'RAD', 'TAX', 'PTRATIO', 'B', 'LSTAT']
x = dataset.data

target = dataset.target
#把label变为(?, 1)维度，为了使用下面的数据集合分割
y = np.reshape(target,(len(target), 1))

#讲数据集1:3比例分割为 测试集：训练集
x_train, x_verify, y_train, y_verify = train_test_split(x, y, random_state=1)

'''
x_train的shape：(379, 13)
y_train的shape：(379, 1)
x_verify的shape：(127, 13)
y_verify 的shape：(127, 1)
'''


'''----------定义线性回归模型，进行训练、预测-----------'''
lr = linear_model.LinearRegression()
lr.fit(x_train,y_train)
y_pred = lr.predict(x_verify)


'''----------图形化预测结果-----------'''
#只显示前50个预测结果，太多的话看起来不直观
plt.xlim([0,50])
plt.plot( range(len(y_verify)), y_verify, 'r', label='y_verify')
plt.plot( range(len(y_pred)), y_pred, 'g--', label='y_predict' )
plt.title('sklearn: Linear Regression')
plt.legend()
plt.savefig('lr/lr-13.png')
plt.show()


'''----------输出模型参数、评价模型-----------'''
print(lr.coef_)
print(lr.intercept_)
print("MSE:",metrics.mean_squared_error(y_verify,y_pred))
print("RMSE:",np.sqrt(metrics.mean_squared_error(y_verify,y_pred)))

#输出模型对应R-Square
print(lr.score(x_train,y_train))
print(lr.score(x_verify,y_verify))

MSE_list=[]
score_train=[]
score_verify=[]

for i in range(2,14):

       x_new = SelectKBest(f_regression, k=i).fit_transform(x, target)
       x_train, x_verify, y_train, y_verify = train_test_split(x_new, y, random_state=1)
       y_train = y_train.reshape(-1)
       train_data = np.insert(x_train, 0, values=y_train, axis=1)

       y_t = train_data[:, 0].reshape(-1,1)
       lr = linear_model.LinearRegression()
       lr.fit(x_train,y_train)
       y_pred = lr.predict(x_verify)

       score_train.append(lr.score(x_train, y_train))
       score_verify.append(lr.score(x_verify, y_verify))
       MSE_list.append(metrics.mean_squared_error(y_verify,y_pred))

print('MSE_list：',MSE_list)
print('train_score:',score_train)
print('verify_score:',score_verify)

plt.subplot(2,1,1)
plt.plot(range(2,14),MSE_list)
plt.title('MSE')
plt.subplot(2,1,2)
plt.plot(range(2,14),score_train,label='score_train')
plt.plot(range(2,14),score_verify,label='score_verify')
plt.legend()
plt.title('R2 score')
plt.savefig('feature_selection.png')
plt.show()