2024/10/13 15:00 linearreg_exercise finish

线性回归练习题
在这个练习中，我们使用一个Kaggle竞赛中提供的共享单车的数据集：Bike Sharing Demand。 该数据集包含2011到2012年Capital
Bikeshare系统中记录的每日每小时单车的租赁数，以及相应的季节和气候等信息。

数据列：

datetime - hourly date + timestamp

season - 1 = spring, 2 = summer, 3 = fall, 4 = winter
holiday - whether the day is considered a holiday
workingday - whether the day is neither a weekend nor holiday
weather - 1: Clear, Few clouds, Partly cloudy, Partly cloudy；2: Mist + Cloudy, Mist + Broken clouds, Mist + Few clouds, Mist；
3: Light Snow, Light Rain + Thunderstorm + Scattered clouds, Light Rain + Scattered clouds；4: Heavy Rain + Ice Pallets +
Thunderstorm + Mist, Snow + Fog
temp - temperature in Celsius
atemp - "feels like" temperature in Celsius
humidity - relative humidity
windspeed - wind speed
casual - number of non-registered user rentals initiated
registered - number of registered user rentals initiated
count - number of total rentals

第一步：读入数据
In [3]: # read the data and set the datetime as the index
import pandas as pd

bikes = pd.read_csv('bikeshare.csv', index_col='datetime', parse_dates=True)

In [4]: bikes.head()

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 1/13

2024/10/13 15:00 linearreg_exercise finish

Out[4]: season holiday workingday weather temp atemp humidity windspeed casual registered count

datetime

2011-01-01 00:00:00 1 0 0 1 9.84 14.395 81 0.0 3 13 16

2011-01-01 01:00:00 1 0 0 1 9.02 13.635 80 0.0 8 32 40

2011-01-01 02:00:00 1 0 0 1 9.02 13.635 80 0.0 5 27 32

2011-01-01 03:00:00 1 0 0 1 9.84 14.395 75 0.0 3 10 13

2011-01-01 04:00:00 1 0 0 1 9.84 14.395 75 0.0 0 1 1

第二步：可视化数据
用matplotlib画出温度“temp”和自行车租赁数“count”之间的散点图；
用seborn画出温度“temp”和自行车租赁数“count”之间带线性关系的散点图（提示：使用seaborn中的lmplot绘制）

In [6]: # matplotlib
import matplotlib.pyplot as plt
# 绘制散点图
plt.figure(figsize=(10, 6))
plt.scatter(bikes['temp'], bikes['count'], alpha=0.5)
plt.grid(True)
plt.show()

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 2/13

2024/10/13 15:00 linearreg_exercise finish

In [8]: # seaborn
import seaborn as sns

# 绘制带线性关系的散点图
plt.figure(figsize=(10, 6))
sns.lmplot(x='temp', y='count', data=bikes, aspect=1.5)
plt.grid(True)
plt.show()

<Figure size 1000x600 with 0 Axes>

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 3/13

2024/10/13 15:00 linearreg_exercise finish

第三步：一元线性回归
用温度预测自行车租赁数

In [9]: # create X and y

X = bikes[['temp']]
y = bikes['count']

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 4/13

2024/10/13 15:00 linearreg_exercise finish

In [10]: # import, instantiate, fit

from sklearn.linear_model import LinearRegression
from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score
import numpy as np
# 实例化线性回归模型
linear_model = LinearRegression()

# 拟合模型
linear_model.fit(X, y)

Out[10]: ▾ LinearRegression i ?

LinearRegression()

In [11]: # print the coefficients

print(f'线性回归模型的系数（slope）: {linear_model.coef_[0]:.4f}')
print(f'线性回归模型的截距（intercept）: {linear_model.intercept_:.4f}')

线性回归模型的系数（slope）: 9.1705
线性回归模型的截距（intercept）: 6.0462

第四步：探索多个特征
In [12]: # explore more features
feature_cols = ['temp', 'season', 'weather', 'humidity']

In [13]: # using seaborn, draw multiple scatter plots between each feature in feature_cols and 'count'
columns_to_plot = feature_cols + ['count']
# 绘制多变量散点图
sns.pairplot(bikes[columns_to_plot], diag_kind='kde')
plt.show()

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 5/13

2024/10/13 15:00 linearreg_exercise finish

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 6/13

2024/10/13 15:00 linearreg_exercise finish

In [14]: # correlation matrix (ranges from 1 to -1)

bikes.corr()

Out[14]: season holiday workingday weather temp atemp humidity windspeed casual registered c

season 1.000000 0.029368 -0.008126 0.008879 0.258689 0.264744 0.190610 -0.147121 0.096758 0.164011 0.16

holiday 0.029368 1.000000 -0.250491 -0.007074 0.000295 -0.005215 0.001929 0.008409 0.043799 -0.020956 -0.00

workingday -0.008126 -0.250491 1.000000 0.033772 0.029966 0.024660 -0.010880 0.013373 -0.319111 0.119460 0.01

weather 0.008879 -0.007074 0.033772 1.000000 -0.055035 -0.055376 0.406244 0.007261 -0.135918 -0.109340 -0.12

temp 0.258689 0.000295 0.029966 -0.055035 1.000000 0.984948 -0.064949 -0.017852 0.467097 0.318571 0.39

atemp 0.264744 -0.005215 0.024660 -0.055376 0.984948 1.000000 -0.043536 -0.057473 0.462067 0.314635 0.38

humidity 0.190610 0.001929 -0.010880 0.406244 -0.064949 -0.043536 1.000000 -0.318607 -0.348187 -0.265458 -0.31

windspeed -0.147121 0.008409 0.013373 0.007261 -0.017852 -0.057473 -0.318607 1.000000 0.092276 0.091052 0.10

casual 0.096758 0.043799 -0.319111 -0.135918 0.467097 0.462067 -0.348187 0.092276 1.000000 0.497250 0.69

registered 0.164011 -0.020956 0.119460 -0.109340 0.318571 0.314635 -0.265458 0.091052 0.497250 1.000000 0.97

count 0.163439 -0.005393 0.011594 -0.128655 0.394454 0.389784 -0.317371 0.101369 0.690414 0.970948 1.00

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 7/13

2024/10/13 15:00 linearreg_exercise finish

In [15]: sns.heatmap(bikes.corr())

Out[15]: <Axes: >

用'temp', 'season', 'weather', 'humidity'四个特征预测单车租赁数'count'

In [23]: # create X and y
# 创建特征矩阵 X 和目标变量 y
feature_cols = ['temp', 'season', 'weather', 'humidity'] # 特征列

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 8/13

2024/10/13 15:00 linearreg_exercise finish

X = bikes[feature_cols] # 特征
y = bikes['count'] # 目标变量

In [17]: # import, instantiate, fit

# 导入必要的库
from sklearn.linear_model import LinearRegression
from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score
# 实例化线性回归模型
linear_model = LinearRegression()

# 拟合模型
linear_model.fit(X, y)

Out[17]: ▾ LinearRegression i ?

LinearRegression()

In [18]: # print the coefficients

print('回归模型的系数:', linear_model.coef_)
print('回归模型的截距:', linear_model.intercept_)

回归模型的系数: [9.17054048]
回归模型的截距: 6.046212959616469

使用train/test split和RMSE来比较多个不同的模型

In [19]: # compare different sets of features

feature_cols1 = ['temp', 'season', 'weather', 'humidity']
feature_cols2 = ['temp', 'season', 'weather']
feature_cols3 = ['temp', 'season', 'humidity']

补充：处理类别特征
有两种类别特征：

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 9/13

2024/10/13 15:00 linearreg_exercise finish

有序类别值： 转换成相应的数字值(例如: small=1, medium=2, large=3)

无序类别值: 使用dummy encoding (0/1编码)

此数据集中的类别特征有：

有序类别值： weather (已经被编码成相应的数字值1,2,3,4)

无序类别值： season (需要进行dummy encoding), holiday (已经被dummy encoded), workingday (已经被dummy encoded)

In [21]: # create dummy variables

season_dummies = pd.get_dummies(bikes.season, prefix='season')

# print 5 random rows

season_dummies.sample(n=5, random_state=1)

Out[21]: season_1 season_2 season_3 season_4

datetime

2011-09-05 11:00:00 False False True False

2012-03-18 04:00:00 True False False False

2012-10-14 17:00:00 False False False True

2011-04-04 15:00:00 False True False False

2012-12-11 02:00:00 False False False True

我们只需要 三个 dummy 变量 (不是四个) （为什么？）, 所以可以删除第一个dummy变量。

In [22]: # drop the first column

season_dummies.drop(season_dummies.columns[0], axis=1, inplace=True)

# print 5 random rows

season_dummies.sample(n=5, random_state=1)

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 10/13

2024/10/13 15:00 linearreg_exercise finish

Out[22]: season_2 season_3 season_4

datetime

2011-09-05 11:00:00 False True False

2012-03-18 04:00:00 False False False

2012-10-14 17:00:00 False False True

2011-04-04 15:00:00 True False False

2012-12-11 02:00:00 False False True

In [23]: # concatenate the original DataFrame and the dummy DataFrame (axis=0 means rows, axis=1 means columns)
bikes = pd.concat([bikes, season_dummies], axis=1)

# print 5 random rows

bikes.sample(n=5, random_state=1)

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 11/13

2024/10/13 15:00 linearreg_exercise finish

Out[23]: season holiday workingday weather temp atemp humidity windspeed casual registered count season_2 season

datetime

2011-
09-05 3 1 0 2 28.70 33.335 74 11.0014 101 207 308 False T
11:00:00

2012-
03-18 1 0 0 2 17.22 21.210 94 11.0014 6 8 14 False Fa
04:00:00

2012-
10-14 4 0 0 1 26.24 31.060 44 12.9980 193 346 539 False Fa
17:00:00

2011-
04-04 2 0 1 1 31.16 33.335 23 36.9974 47 96 143 True Fa
15:00:00

2012-
12-11 4 0 1 2 16.40 20.455 66 22.0028 0 1 1 False Fa
02:00:00

将编码成的dummy变量加入回归模型的特征，预测单车租赁数
In [30]: # include dummy variables for season in the model
feature_cols = ['temp', 'season_2', 'season_3', 'season_4', 'humidity']
X = bikes[feature_cols] # 特征
y = bikes['count'] # 目标变量

# 划分训练集和测试集 (70%训练，30%测试)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)

# 实例化线性回归模型
linear_model = LinearRegression()

# 拟合模型

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 12/13

2024/10/13 15:00 linearreg_exercise finish

linear_model.fit(X_train, y_train)

# 打印模型的系数和截距
print('回归模型的系数:', linear_model.coef_)
print('回归模型的截距:', linear_model.intercept_)

# 使用测试集进行预测
y_pred = linear_model.predict(X_test)

# 计算模型的R²评分
r2 = r2_score(y_test, y_pred)
print(f'新模型的 R² 评分: {r2:.4f}')

回归模型的系数: [ 11.25840607 -5.55599206 -46.71747057 67.71409193 -2.80624827]

回归模型的截距: 133.0147787618583
新模型的 R² 评分: 0.2706

和前面的模型进行比较

In [ ]:

localhost:8888/lab/tree/linearreg_exercise finish.ipynb 13/13