如何将 Scikit-learn Python 库用于数据科学项目
灵活多样的 Python 库为数据分析和数据挖掘提供了强力的机器学习工具。
Scikit-learn Python 库最初于 2007 年发布,通常用于解决各种方面的机器学习和数据科学问题。这个多种功能的库提供了整洁、一致、高效的 API 和全面的在线文档。
什么是 Scikit-learn?
Scikit-learn 是一个开源 Python 库,拥有强大的数据分析和数据挖掘工具。 在 BSD 许可下可用,并建立在以下机器学习库上:
NumPy
,一个用于操作多维数组和矩阵的库。它还具有广泛的数学函数汇集,可用于执行各种计算。SciPy
,一个由各种库组成的生态系统,用于完成技术计算任务。Matplotlib
,一个用于绘制各种图表和图形的库。
Scikit-learn 提供了广泛的内置算法,可以充分用于数据科学项目。
以下是使用 Scikit-learn 库的主要方法。
1、分类
分类工具识别与提供的数据相关联的类别。例如,它们可用于将电子邮件分类为垃圾邮件或非垃圾邮件。
Scikit-learn 中的分类算法包括:
- 支持向量机Support vector machines(SVM)
- 最邻近Nearest neighbors
- 随机森林Random forest
2、回归
回归涉及到创建一个模型去试图理解输入和输出数据之间的关系。例如,回归工具可用于理解股票价格的行为。
回归算法包括:
- 支持向量机Support vector machines(SVM)
- 岭回归Ridge regression
- Lasso(LCTT 译注:Lasso 即 least absolute shrinkage and selection operator,又译为最小绝对值收敛和选择算子、套索算法)
3、聚类
Scikit-learn 聚类工具用于自动将具有相同特征的数据分组。 例如,可以根据客户数据的地点对客户数据进行细分。
聚类算法包括:
- K-means
- 谱聚类Spectral clustering
- Mean-shift
4、降维
降维降低了用于分析的随机变量的数量。例如,为了提高可视化效率,可能不会考虑外围数据。
降维算法包括:
- 主成分分析Principal component analysis(PCA)
- 功能选择Feature selection
- 非负矩阵分解Non-negative matrix factorization
5、模型选择
模型选择算法提供了用于比较、验证和选择要在数据科学项目中使用的最佳参数和模型的工具。
通过参数调整能够增强精度的模型选择模块包括:
- 网格搜索Grid search
- 交叉验证Cross-validation
- 指标Metrics
6、预处理
Scikit-learn 预处理工具在数据分析期间的特征提取和规范化中非常重要。 例如,您可以使用这些工具转换输入数据(如文本)并在分析中应用其特征。
预处理模块包括:
- 预处理
- 特征提取
Scikit-learn 库示例
让我们用一个简单的例子来说明如何在数据科学项目中使用 Scikit-learn 库。
我们将使用鸢尾花花卉数据集,该数据集包含在 Scikit-learn 库中。 鸢尾花数据集包含有关三种花种的 150 个细节,三种花种分别为:
- Setosa:标记为 0
- Versicolor:标记为 1
- Virginica:标记为 2
数据集包括每种花种的以下特征(以厘米为单位):
- 萼片长度
- 萼片宽度
- 花瓣长度
- 花瓣宽度
第 1 步:导入库
由于鸢尾花花卉数据集包含在 Scikit-learn 数据科学库中,我们可以将其加载到我们的工作区中,如下所示:
<span class="kwd">from</span><span class="pln"> sklearn </span><span class="kwd">import</span><span class="pln"> datasets</span>
<span class="pln">iris </span><span class="pun">=</span><span class="pln"> datasets</span><span class="pun">.</span><span class="pln">load_iris</span><span class="pun">()</span>
这些命令从 sklearn
导入数据集 datasets
模块,然后使用 datasets
中的 load_iris()
方法将数据包含在工作空间中。
第 2 步:获取数据集特征
数据集 datasets
模块包含几种方法,使您更容易熟悉处理数据。
在 Scikit-learn 中,数据集指的是类似字典的对象,其中包含有关数据的所有详细信息。 使用 .data
键存储数据,该数据列是一个数组列表。
例如,我们可以利用 iris.data
输出有关鸢尾花花卉数据集的信息。
<span class="kwd">print</span><span class="pun">(</span><span class="pln">iris</span><span class="pun">.</span><span class="pln">data</span><span class="pun">)</span>
这是输出(结果已被截断):
<span class="pun">[[</span><span class="lit">5.1</span><span class="lit">3.5</span><span class="lit">1.4</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.9</span><span class="lit">3.</span><span class="pln"> </span><span class="lit">1.4</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.7</span><span class="lit">3.2</span><span class="lit">1.3</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.6</span><span class="lit">3.1</span><span class="lit">1.5</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">5.</span><span class="pln"> </span><span class="lit">3.6</span><span class="lit">1.4</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">5.4</span><span class="lit">3.9</span><span class="lit">1.7</span><span class="lit">0.4</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.6</span><span class="lit">3.4</span><span class="lit">1.4</span><span class="lit">0.3</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">5.</span><span class="pln"> </span><span class="lit">3.4</span><span class="lit">1.5</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.4</span><span class="lit">2.9</span><span class="lit">1.4</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.9</span><span class="lit">3.1</span><span class="lit">1.5</span><span class="lit">0.1</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">5.4</span><span class="lit">3.7</span><span class="lit">1.5</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.8</span><span class="lit">3.4</span><span class="lit">1.6</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.8</span><span class="lit">3.</span><span class="pln"> </span><span class="lit">1.4</span><span class="lit">0.1</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">4.3</span><span class="lit">3.</span><span class="pln"> </span><span class="lit">1.1</span><span class="lit">0.1</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">5.8</span><span class="lit">4.</span><span class="pln"> </span><span class="lit">1.2</span><span class="lit">0.2</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">5.7</span><span class="lit">4.4</span><span class="lit">1.5</span><span class="lit">0.4</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">5.4</span><span class="lit">3.9</span><span class="lit">1.3</span><span class="lit">0.4</span><span class="pun">]</span>
<span class="pln"> </span><span class="pun">[</span><span class="lit">5.1</span><span class="lit">3.5</span><span class="lit">1.4</span><span class="lit">0.3</span><span class="pun">]</span>
我们还使用 iris.target
向我们提供有关花朵不同标签的信息。
<span class="kwd">print</span><span class="pun">(</span><span class="pln">iris</span><span class="pun">.</span><span class="pln">target</span><span class="pun">)</span>
这是输出:
<span class="pun">[</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span>
<span class="pln"> </span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">0</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span>
<span class="pln"> </span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">1</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span>
<span class="pln"> </span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span><span class="lit">2</span>
<span class="pln"> </span><span class="lit">2</span><span class="lit">2</span><span class="pun">]</span>
如果我们使用 iris.target_names
,我们将输出数据集中找到的标签名称的数组。
<span class="kwd">print</span><span class="pun">(</span><span class="pln">iris</span><span class="pun">.</span><span class="pln">target_names</span><span class="pun">)</span>
以下是运行 Python 代码后的结果:
<span class="pun">[</span><span class="str">'setosa'</span><span class="str">'versicolor'</span><span class="str">'virginica'</span><span class="pun">]</span>
第 3 步:可视化数据集
我们可以使用箱形图来生成鸢尾花数据集的视觉描绘。 箱形图说明了数据如何通过四分位数在平面上分布的。
以下是如何实现这一目标:
<span class="kwd">import</span><span class="pln"> seaborn </span><span class="kwd">as</span><span class="pln"> sns</span>
<span class="pln">box_data </span><span class="pun">=</span><span class="pln"> iris</span><span class="pun">.</span><span class="pln">data </span><span class="com">#</span><span class="pun">表示数据数组的变量</span>
<span class="pln">box_target </span><span class="pun">=</span><span class="pln"> iris</span><span class="pun">.</span><span class="pln">target </span><span class="com">#</span><span class="pun">表示标签数组的变量</span>
<span class="pln">sns</span><span class="pun">.</span><span class="pln">boxplot</span><span class="pun">(</span><span class="pln">data </span><span class="pun">=</span><span class="pln"> box_data</span><span class="pun">,</span><span class="pln">width</span><span class="pun">=</span><span class="lit">0.5</span><span class="pun">,</span><span class="pln">fliersize</span><span class="pun">=</span><span class="lit">5</span><span class="pun">)</span>
<span class="pln">sns</span><span class="pun">.</span><span class="kwd">set</span><span class="pun">(</span><span class="pln">rc</span><span class="pun">={</span><span class="str">'figure.figsize'</span><span class="pun">:(</span><span class="lit">2</span><span class="pun">,</span><span class="lit">15</span><span class="pun">)})</span>
让我们看看结果:
在横轴上:
- 0 是萼片长度
- 1 是萼片宽度
- 2 是花瓣长度
- 3 是花瓣宽度
垂直轴的尺寸以厘米为单位。
总结
以下是这个简单的 Scikit-learn 数据科学教程的完整代码。
<span class="kwd">from</span><span class="pln"> sklearn </span><span class="kwd">import</span><span class="pln"> datasets</span>
<span class="pln">iris </span><span class="pun">=</span><span class="pln"> datasets</span><span class="pun">.</span><span class="pln">load_iris</span><span class="pun">()</span>
<span class="kwd">print</span><span class="pun">(</span><span class="pln">iris</span><span class="pun">.</span><span class="pln">data</span><span class="pun">)</span>
<span class="kwd">print</span><span class="pun">(</span><span class="pln">iris</span><span class="pun">.</span><span class="pln">target</span><span class="pun">)</span>
<span class="kwd">print</span><span class="pun">(</span><span class="pln">iris</span><span class="pun">.</span><span class="pln">target_names</span><span class="pun">)</span>
<span class="kwd">import</span><span class="pln"> seaborn </span><span class="kwd">as</span><span class="pln"> sns</span>
<span class="pln">box_data </span><span class="pun">=</span><span class="pln"> iris</span><span class="pun">.</span><span class="pln">data </span><span class="com">#</span><span class="pun">表示数据数组的变量</span>
<span class="pln">box_target </span><span class="pun">=</span><span class="pln"> iris</span><span class="pun">.</span><span class="pln">target </span><span class="com">#</span><span class="pun">表示标签数组的变量</span>
<span class="pln">sns</span><span class="pun">.</span><span class="pln">boxplot</span><span class="pun">(</span><span class="pln">data </span><span class="pun">=</span><span class="pln"> box_data</span><span class="pun">,</span><span class="pln">width</span><span class="pun">=</span><span class="lit">0.5</span><span class="pun">,</span><span class="pln">fliersize</span><span class="pun">=</span><span class="lit">5</span><span class="pun">)</span>
<span class="pln">sns</span><span class="pun">.</span><span class="kwd">set</span><span class="pun">(</span><span class="pln">rc</span><span class="pun">={</span><span class="str">'figure.figsize'</span><span class="pun">:(</span><span class="lit">2</span><span class="pun">,</span><span class="lit">15</span><span class="pun">)})</span>
Scikit-learn 是一个多功能的 Python 库,可用于高效完成数据科学项目。
如果您想了解更多信息,请查看 LiveEdu 上的教程,例如 Andrey Bulezyuk 关于使用 Scikit-learn 库创建机器学习应用程序的视频。
有什么评价或者疑问吗? 欢迎在下面分享。
作者:Dr.Michael J.Garbade 选题:lujun9972 译者:Flowsnow 校对:wxy