【Python】线性回归模型

2021 年 09 月 27 日

1691 次浏览

2440字数

- 数据准备

```python
import numpy as np
from sklearn import linear_model
import matplotlib.pyplot as plt
# 数据准备
x = np.array([1,3,2,1,3])
y = np.array([14,24,18,17,27])

regr = linear_model.LinearRegression()
# 模型训练
regr.fit(x.reshape(-1,1),y)
```

- 数据处理

```python
# 得到直线的斜率、截距、公式 y = a*x + b
a,b = regr.coef_,regr.intercept_
print('a=',a,'b=',b)

# 对回归模型打分
print('score=',regr.score(x.reshape(-1,1),y))

# 可视化模型
# 1.真实的点
plt.scatter(x,y,color='blue')
# 2.拟合的直线
plt.plot(x,regr.predict(x.reshape(-1,1)),color='red',linewidth=4)
plt.show()

# 对数据进行预测
c = np.array([6])
print('y_pre=',regr.predict(c.reshape(-1,1)))
```

### 多元线性回归

```python
from sklearn import datasets # 从scikit-learn导入数据集
load_data = datasets.load_boston() # 从数据集库加载波士顿数集

import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split

#导入数据
data_x = load_data.data
data_y = load_data.target
x_train,x_verify,y_train,y_verify = train_test_split(data_x,data_y,train_size=0.8,test_size=0.2)
```

```python
# 建立线性回归模型及训练
lm = linear_model.LinearRegression()
lm.fit(x_train,y_train)
# 得到直线的斜率、截距、公式 y=a*x+b
a,b = lm.coef_,lm.intercept_
print('a=',a,'\nb=',b)
# 对回归模型进行检查，给实际观测点和回归方程的拟合程度打分
print('train_score=',lm.score(x_train,y_train))
# 对新输一局进行预测
predictions = lm.predict(x_verify)
print('test_score=',lm.score(x_verify,y_verify))
```

### 全连接神经网络回归

```python
from sklearn.neural_network import MLPRegressor
from sklearn.preprocessing import StandardScaler
from sklearn import datasets #从scikit-learn导入数据集
load_data = datasets.load_boston() # 从数据集库加载波士顿数据集
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
#导入数据
data_x = load_data.data
data_y = load_data.target
scaler = StandardScaler() # 标准化转换
scaler.fit(data_x) # 训练标准化对象
data_x = scaler.transform(data_x) # 转换数据集
x_train,x_verify,y_train,y_verify=train_test_split(data_x,data_y,train_size=0.8,test_size=0.2)

# 神经网络回归
clf = MLPRegressor(solver='lbfgs',alpha=1e-5,hidden_layer_sizes=(5,2),random_state=1)
clf.fit(x_train,y_train)
# 查看训练模型得分
print('train_score=',clf.score(x_train,y_train))
# 模型预测
pred = clf.predict(x_verify)
# 查看预测得分
print('test_score=',clf.score(x_verify,y_verify))
```

```python
# 查看网络结构
index=0
for w in clf.coefs_:
    index += 1
    print('第{}层网络层：'.format(index))
    print('权重矩阵大小：',w.shape)
    print('系数矩阵：',w)
```

最后修改：2021 年 09 月 27 日