机器学习好伙伴之scikit-learn的使用——datasets获得数据集

在上一节课我们说到如何进行训练集和测试集的划分，在划分之前，还有一点很重要的是载入合适的数据集。

scikit-learn自带许多实用的数据集，用于测试模型与训练，非常实用

载入sklearn中自带的datesets

利用如下函数可以载入sklearn中常用的datesets

from sklearn import datasets

在datasets中，常用的数据名称、调用方式、适用算法和数据规模如下：

除去上面的几种数据集外，我们可以前往sklearn的官网查看其它实用的数据集：
https://scikit-learn.org/stable/modules/classes.html#module-sklearn.datasets

在python代码中，可以通过如下的方法获取数据集。

loaded_data = datasets.load_boston()
# 获取特征与对应的目标值
data_X = loaded_data.data
data_Y = loaded_data.target

利用sklearn的函数生成数据

除去sklearn中自带的许多数据集外，sklearn还可以生成数据。
常用的生成方法如下：

本文以生成回归数据为例，讲解如何利用sklearn生成可以用于训练的数据。
打开datasets.make_regression的页面可以看到datasets.make_regression详细的参数。

常用参数如下：
n_samples：样本数
n_features：每个样本的特征数
n_informative ：有效信息，信息特征的数量，即用于构建用于生成输出的线性模型的特征的数量。
n_redundant：冗余信息，informative特征的随机线性组合
n_repeated ：重复信息，随机提取n_informative和n_redundant 特征
n_targets：回归目标的输出
bias：基本线性模型中的偏差项
noise：生成结果的噪声

函数调用方式为：

X, y = datasets.make_regression(n_samples=100, n_features=1, n_targets=1, noise=10)

应用示例

利用sklearn中自带的datesets进行训练

实现代码为：

# 载入数据集
from sklearn import datasets
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
import numpy as np

# 获取自带的数据库
loaded_data = datasets.load_boston()
data_X = loaded_data.data
data_Y = loaded_data.target

# 对自带的数据库进行划分
X_train, X_test, y_train, y_test = train_test_split(data_X, data_Y, test_size=0.3)

# 进行线性训练
model = LinearRegression()
model.fit(X_train, y_train)

# 预测，比较结果
print(model.predict(X_test)-y_test)

实验结果为，预测结果与实际结果有一定的偏差：

[ -4.32083993   1.21779322   0.92588852   6.63213006   3.64206762
   0.53339736   0.58771424  -2.57979413   1.24509333   6.5652264
   6.96686417   4.44253819  -0.80383831  -2.65691237  -6.65582548
   2.51880244 -20.26150608  -5.07295788   3.14756883   1.79616885
   1.00446839   0.95839837   6.98646141   0.67669195   2.27505619
  -5.2557447    5.31508681  -1.78250561   1.52031694   0.95947612
   0.36195507  -0.9259978   -1.36718779   4.27230508   0.83577059
  -2.70704876   2.5626411   -0.96554518  -2.08485853   4.52656586
  -0.56106616  -0.82236156   0.32942509   2.87786991  -1.57245813
   3.56041759  -0.89990552  -0.70613626   0.56616644  -0.44223704
  -2.96971507   1.59152775 -27.33888072   2.85800776   0.62582613
   1.64704949 -10.38681447  -0.20579836  -5.12782407  -1.95571152
   1.46984184   6.26040771   3.61214832  -5.15636024   3.57058735
   0.69584653   0.04508708   2.39937335  -3.4427129   -7.23339761
 -13.25181432   5.32245934   4.19778417  -3.38290778   1.17488126
   2.88578286  -2.46635697   2.9071728    1.99747009   3.62120858
   2.80418927   4.96581854   6.89986873   5.25073363  -2.66358202
   6.16654885   4.70551011  -2.50297777   0.29701667   2.30451103
 -18.69710415   2.44340239   5.0921586    2.07129662  -9.12512144
  -7.23221737   3.06076078   3.2289587    0.57908266  -6.66312125
   3.36308189   0.15555137   1.10987299   1.20843313  -3.08905426
   4.62481519  -2.97478583   3.44949672   0.62329198   2.05383104
   0.87283797   1.89363157  -9.48663782  -0.04938579   1.13795316
  -1.93810385   2.16260112  -4.61696882   1.37781137  -1.77504111
   0.39900168   2.95068343  -0.05496372  -1.04251617   6.67036798
 -11.42967859  -1.03175378   0.95026778   6.29436769  -8.19767754
  -1.61315791  -1.19523958  -2.27840324  -2.81928986  -2.64508748
   1.28006411 -30.939282     0.9696373   -0.87861839  15.39549095
   0.93702141  -0.52928773   0.97060633 -18.11069675   1.94384285
   6.0846876   -0.32219858   1.6885332    1.72306516   0.30902564
   1.1483849    1.69583331]

利用sklearn中生成的数据进行训练

实现代码为：

# 载入数据集
from sklearn import datasets
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
import numpy as np

# 利用函数获取测试内容
data_X, data_Y = datasets.make_regression(n_samples=500, n_features=4, n_targets=1, noise=10)

# 对内容进行划分
X_train, X_test, y_train, y_test = train_test_split(data_X, data_Y, test_size=0.3)

# 进行线性训练
model = LinearRegression()
model.fit(X_train, y_train)

# 预测，比较结果
print(model.predict(X_test)-y_test)

实验结果为，预测结果与实际结果有一定的偏差：

[ 18.02467269  10.65427027   3.6119628   -2.31792705  17.12008342
   4.78838342  15.18861015 -26.5113201    1.89947597   7.82148113
 -10.10536587  -5.29198345   5.77718464   1.03981516  10.97207192
  10.16088225  -4.44089219 -11.55399703  -9.03323276   0.90299202
  -7.89203736   6.11155059 -22.18435471   8.71301653   3.36389708
 -10.76470998 -24.07692007  -4.46077878   1.03771658  -5.39892106
   5.72751355   4.95228045   6.6005534    0.1526285   -7.07137473
   8.9840744    5.07088902  -3.62815639   1.86320243   4.42907918
   8.04422199  -7.58088343   0.3912087    5.27823853  -8.03661081
  -4.57543018 -10.0447772   -1.7485529    9.56039568   2.05514496
   0.4758245   -8.4528048   -1.34800374  11.7522642    0.38639048
   0.76524824   5.63207193 -10.98782417  27.91112702  -5.80363586
 -12.70346101   9.14590667  -1.46200611   9.97969014   9.96405817
   1.38061919  -5.35214404  -4.41259465  12.11166841  12.97697531
   2.8912008    8.62946742  -0.04821456  -6.50127461  11.24254809
   4.68218756 -33.56564161  -9.57347681  -9.17343223  11.77168043
   3.35652864  -8.45436188  18.56012243  -3.14858898   9.59542618
   2.42877687  -5.48526428  17.01071347  -2.14069015   4.39177413
   5.57498895  -2.38403742  -2.43446519  14.36727563  -0.88722033
  10.04797528  -9.33955844  -3.68504439   6.69969735 -29.4115413
  -1.21425655   1.51859444   6.08471639  14.09008583  16.08143523
  13.34213548  -4.66548477  -4.12516003  -2.4253649   -3.39698823
  -2.29500693  -0.12027219  -0.23804341 -17.72229705   2.10578861
  -3.99810849 -14.97580105   1.05624791   5.37911996  -3.46126258
  -0.32871116  -4.99624719  10.59211956  14.24566851   5.71547039
   4.52283305  -3.10016563 -15.05354972  -5.74515295  12.51473663
   4.63068615  -3.20664498  -5.42049079   9.37493256   2.2296205
   1.30180444   6.47528801  -9.26910871   5.36695885  -5.05384241
 -15.74364437 -13.61850084 -22.27715077  12.15521687  -1.81656035
  -3.71082241  -1.73377178  -8.61096813   6.1261822    4.22906206]