目录
1 下载数据集和预处理
1.1 加载/下载数据集
1.2 数据可视化
1.3 数据清洗
1.4 特征工程
1.5 构建特征集和标签集
1.6 拆分训练集和测试集
2 训练模型
2.1 选择算法和确定模型
2.2 训练拟合模型
3 评估并优化模型性能
本文以糖尿病数据集diabetes为基础进行线性回归训练：
1 下载数据集和预处理
1.1 加载/下载数据集
"""
@Title: 收集数据
@Time: 2024/3/11
@Author: Michael Jie
收集数据和预处理：
1、收集数据；
2、数据可视化；
3、数据清洗；
4、特征工程；
5、构建特征集和标签集（仅监督学习需要）；
6、拆分训练集和测试集。
"""
import sklearn.datasets as ds
import pandas as pd
# 加载并返回糖尿病数据集（回归）
diabetes = ds.load_diabetes(
# 若为True，返回(data, target)元组，而非Bunch对象
return_X_y=False,
# 若为True，以pandas DataFrame/Series形式返回数据集
as_frame=False,
# 若为True，返回归一化后的特征集
scaled=False
)
# Bunch对象本质是一个字典
print(diabetes.keys())
"""
dict_keys([
'data', # 特征集
'target', # 标签集
'frame', # 包含特征值和标签的数组，当as_frame=True时存在
'DESCR', # 数据集描述
'feature_names', # 特征集列名
'data_filename', # 内存中的特征集文件名
'target_filename', # 内存中的标签集文件名
'data_module'
])
"""
# 特征集
data = diabetes.data
print(type(data), data.shape)
"""
<class 'numpy.ndarray'>
(442, 10)
"""
feature_names = diabetes.feature_names
print(feature_names, type(feature_names))
"""
['age', 'sex', 'bmi', 'bp', 's1', 's2', 's3', 's4', 's5', 's6']
<class 'list'>
"""
# 标签集
target = diabetes.target
print(type(target), target.shape)
"""
<class 'numpy.ndarray'>
(442,)
"""
# 数据集描述
print(diabetes.DESCR)
"""
Diabetes dataset
----------------
Ten baseline variables, age, sex, body mass index, average blood
pressure, and six blood serum measurements were obtained for each of n =
442 diabetes patients, as well as the response of interest, a
quantitative measure of disease progression one year after baseline.
**Data Set Characteristics:**
:Number of Instances: 442
:Number of Attributes: First 10 columns are numeric predictive values
:Target: Column 11 is a quantitative measure of disease progression one year after baseline
:Attribute Information:
- age age in years
- sex
- bmi body mass index
- bp average blood pressure
- s1 tc, total serum cholesterol
- s2 ldl, low-density lipoproteins
- s3 hdl, high-density lipoproteins
- s4 tch, total cholesterol / HDL
- s5 ltg, possibly log of serum triglycerides level
- s6 glu, blood sugar level
Note: Each of these 10 feature variables have been mean centered and scaled by the standard deviation times the square root of `n_samples` (i.e. the sum of squares of each column totals 1).
Source URL:
https://www4.stat.ncsu.edu/~boos/var.select/diabetes.html
For more information see:
Bradley Efron, Trevor Hastie, Iain Johnstone and Robert Tibshirani (2004) "Least Angle Regression," Annals of Statistics (with discussion), 407-499.
(https://web.stanford.edu/~hastie/Papers/LARS/LeastAngle_2002.pdf)
"""
# 下载数据集
data_csv = pd.DataFrame(data=data, columns=feature_names)
target_csv = pd.DataFrame(data=target, columns=['target'])
diabetes_csv = pd.concat([data_csv, target_csv], axis=1)
diabetes_csv.to_csv(r'diabetes_datasets.csv', index=False)
1.2 数据可视化
"""
@Title: 数据可视化
@Time: 2024/3/11
@Author: Michael Jie
"""
import pandas as pd
import matplotlib.pyplot as plt
# 读取数据
csv = pd.read_csv(r'diabetes_datasets.csv')
print(csv.shape) # (442, 11)
# 可视化数据
plt.figure(figsize=(19.2, 10.8))
for i in range(csv.shape[1] - 1):
plt.subplot(2, 5, i + 1).scatter(csv[csv.columns[i]], csv["target"])
# 保存图片
plt.savefig(r'diabetes_datasets.png')
# plt.show()
1.3 数据清洗
"""
@Title: 数据清洗
@Time: 2024/3/11
@Author: Michael Jie
"""
import pandas as pd
"""
1、处理缺失数据：剔除残缺数据，也可以用平均值、随机值或者0来补值；
2、处理重复数据：删除完全相同的重复数据处理；
3、处理错误数据：处理逻辑错误数据；
4、处理不可用数据：处理格式错误数据。
"""
# 读取数据
csv = pd.read_csv(r'diabetes_datasets.csv')
# 统计NaN出现的次数
print(csv.isna().sum())
"""
age 0
sex 0
bmi 0
bp 0
s1 0
s2 0
s3 0
s4 0
s5 0
s6 0
target 0
dtype: int64
"""
1.4 特征工程
"""
@Title: 特征工程
@Time: 2024/3/11
@Author: Michael Jie
"""
import numpy as np
import sklearn.datasets as ds
# 标准化
def z_score_normalization(x, axis=0):
x = np.array(x)
x = (x - np.mean(x, axis=axis)) / np.std(x, axis=axis)
return x
# 若为True，返回归一化后的特征集
diabetes_pre = ds.load_diabetes(scaled=True)
print(diabetes_pre.data)
# 手动标准化特征集
diabetes = ds.load_diabetes(scaled=False)
print(z_score_normalization(diabetes.data))
1.5 构建特征集和标签集
无。
1.6 拆分训练集和测试集
"""
@Title:
@Time: 2024/3/11
@Author: Michael Jie
"""
import sklearn.datasets as ds
from sklearn.model_selection import train_test_split
# 加载数据
diabetes = ds.load_diabetes(scaled=False)
# 将数据集进行80%训练集和20%的测试集的分割
x_train, x_test, y_train, y_test = train_test_split(
diabetes.data, diabetes.target, test_size=0.2, random_state=0
)
print(x_train.shape, x_test.shape, y_train.shape, y_test.shape)
"""
(353, 10) (89, 10) (353,) (89,)
"""
2 训练模型
2.1 选择算法和确定模型
# 创建基本线性回归类
linear = LinearRegression(
# 是否计算截距
fit_intercept=True,
# 是否拷贝特征集
copy_X=True,
)
# 创建正则线性回归类
ridge = Ridge(
# 学习率
alpha=1.0,
# 是否计算截距
fit_intercept=True,
# 是否拷贝特征集
copy_X=True,
# 最大训练轮次
max_iter=None,
# 最小损失差
tol=1e-4,
)2.2 训练拟合模型
"""
@Title: 训练模型和评估
@Time: 2024/3/11
@Author: Michael Jie
"""
import sklearn.datasets as ds
from sklearn.linear_model import LinearRegression, Ridge
from sklearn.model_selection import train_test_split
# 加载数据
diabetes = ds.load_diabetes(scaled=True)
# 将数据集进行80%的训练集和20%的测试集的分割
x_train, x_test, y_train, y_test = train_test_split(
diabetes.data, diabetes.target, test_size=0.2, random_state=0
)
# 创建基本线性回归类
linear = LinearRegression()
# 训练
linear.fit(x_train, y_train)
print(linear.coef_, linear.intercept_)
"""
[ -35.55025079 -243.16508959 562.76234744 305.46348218 -662.70290089
324.20738537 24.74879489 170.3249615 731.63743545 43.0309307 ] 152.5380470138517
"""
# 创建正则线性回归类
ridge = Ridge()
# 训练
ridge.fit(x_train, y_train)
print(ridge.coef_, ridge.intercept_)
"""
[ 21.34794489 -72.97401935 301.36593604 177.49036347 2.82093648
-35.27784862 -155.52090285 118.33395129 257.37783937 102.22540041] 151.9441509473086
"""
3 评估并优化模型性能
# 创建基本线性回归类
linear = LinearRegression()
linear.fit(x_train, y_train)
# 评估模型，结果在0-1之间，越大证明模型越拟合数据
print(linear.score(x_test, y_test))
"""
0.33223321731061806
"""
# 创建正则线性回归类
ridge = Ridge()
ridge.fit(x_train, y_train)
# 评估模型
print(ridge.score(x_test, y_test))
"""
0.3409800318493461
"""
