Python机器学习实践项目1——线性回归预测汽车油耗里程数_使用局部线性回归模型预测汽车的燃油效率

1. 导入python库和数据

import pandas as pd
import matplotlib
import sklearn
from sklearn.linear_model import LinearRegression
import matplotlib.pyplot as plt
 
for i in [pd, matplotlib, sklearn]:
    print(i.__name__,": ",i.__version__, sep="")

输出：

pandas: 0.25.3
matplotlib: 3.1.2
sklearn: 0.21.3

2. 导入数据

columns = ["mpg", "cylinders", "displacement", "horsepower", "weight", "acceleration", "model year", "origin", "car name"]
cars = pd.read_table("auto-mpg.data", delim_whitespace=True, names=columns)
print(cars.head(5))
print(cars.dtypes)

输出：

    mpg  cylinders  displacement  horsepower  weight  acceleration  \
0  18.0          8         307.0       130.0  3504.0          12.0   
1  15.0          8         350.0       165.0  3693.0          11.5   
2  18.0          8         318.0       150.0  3436.0          11.0   
3  16.0          8         304.0       150.0  3433.0          12.0   
4  17.0          8         302.0       140.0  3449.0          10.5   
 
   model year  origin                   car name  
0          70       1  chevrolet chevelle malibu  
1          70       1          buick skylark 320  
2          70       1         plymouth satellite  
3          70       1              amc rebel sst  
4          70       1                ford torino  
mpg             float64
cylinders         int64
displacement    float64
horsepower      float64
weight          float64
acceleration    float64
model year        int64
origin            int64
car name         object
dtype: object

3. 观察数据特征之间的关系

import matplotlib.gridspec as gridspec
fig = plt.figure(figsize=(10,10),tight_layout=True)
 
#gs = gridspec.GridSpec(3,3)
 
for i,item in enumerate(["cylinders", "displacement", "horsepower", "weight", "acceleration", \
          "model year", "origin"],start=1):
    ax = fig.add_subplot(3,3,i)
    ax.scatter(cars[item],cars["mpg"])
    ax.set_xlabel(item)
    ax.set_ylabel("mpg")
    
plt.show()

输出：

从如上结果可以发现，displacement, horsepower, weight三个指标与mpg有比较明显的相关性，acceleration与mpg相关性较差，其余指标均是离散化的，无法很好地展示出与mpg的相关性。

4. 训练模型与评估

4.1 weight与mpg

选取汽车重量（weight)指标和每加仑油耗（mpg）进行模型预测

lr = LinearRegression(fit_intercept=True) # 模型初始化
lr.fit(cars[["weight"]], cars["mpg"]) # 训练
 
predictions = lr.predict(cars[["weight"]]) # 预测
print(predictions[0:5])
print(cars["mpg"][0:5])

输出：

[19.41852276 17.96764345 19.94053224 19.96356207 19.84073631]
0    18.0
1    15.0
2    18.0
3    16.0
4    17.0
Name: mpg, dtype: float64

plt.scatter(cars["weight"], cars["mpg"], c='red', label="mpg")
plt.scatter(cars["weight"], predictions, c='blue', label="predictions")
plt.legend()  # 加上legend
plt.show()

输出：

模型评估

lr = LinearRegression()
lr.fit(cars[["weight"]], cars["mpg"])
predictions = lr.predict(cars[["weight"]])
 
from sklearn.metrics import mean_squared_error  # 导入均方误差函数
mse = mean_squared_error(cars["mpg"], predictions)
print(mse)

输出：

18.780939734628394

mse = mean_squared_error(cars["mpg"], predictions)
rmse = mse ** (0.5)  # 对均方误差开根号
print (rmse)

输出：

4.333698159150957

4.2 horsepower与mpg

lr = LinearRegression(fit_intercept=True) # 模型初始化
lr.fit(cars[["horsepower"]], cars["mpg"]) # 训练
 
predictions = lr.predict(cars[["horsepower"]]) # 预测
print(predictions[0:5])
print(cars["mpg"][0:5])

输出：

[19.41604569 13.89148002 16.25915102 16.25915102 17.83759835]
0    18.0
1    15.0
2    18.0
3    16.0
4    17.0
Name: mpg, dtype: float64

plt.scatter(cars["horsepower"], cars["mpg"], c='red', label="mpg")
plt.scatter(cars["horsepower"], predictions, c='blue', label="predictions")
plt.legend()  # 加上legend
plt.show()

输出：

from sklearn.metrics import mean_squared_error
 
lr = LinearRegression()
lr.fit(cars[["horsepower"]], cars["mpg"])
predictions = lr.predict(cars[["horsepower"]])
 
 
mse = mean_squared_error(cars["mpg"], predictions)
rmse = mse ** (0.5)  # 对均方误差开根号
print (rmse)

输出：

4.893226230065713

4.3 其它指标与mpg

可参照上面的例子自行对其它指标进行分析