高斯朴素贝叶斯方法进行鸢尾花分类

骑猪看日落 2022-04-16 05:21 293阅读 0赞

### 贝叶斯方法完整代码 ###

import seaborn as sns  
    iris = sns.load_dataset('iris')
    
    X_iris = iris.drop('species', axis=1)
    y_iris = iris['species']
    
    print(X_iris)
    from sklearn.cross_validation import train_test_split
    Xtrain, Xtest, ytrain, ytest = train_test_split(X_iris, y_iris, random_state=1)
    
    from sklearn.naive_bayes import GaussianNB
    model = GaussianNB()
    model.fit(Xtrain, ytrain)
    y_model = model.predict(Xtest)
    
    from sklearn.metrics import accuracy_score
    accuracy_score(ytest, y_model)

--------------------

## 步骤分析 ##

### 一-首先获取数据. ###

这里我们在线导入seaborn库的iris(鸢尾花)数据

import seaborn as sns
    iris = sns.load_dataset('iris')

[这是github上的说明][github], 可直接下载csv文件

### 二-将数据格式化 ###

X_iris = iris.drop('species', axis=1)
    y_iris = iris['species']

将类别那一列删除生成新的对象赋值给X\_iris, y\_iris为分类.  
[pandas的drop方法参考][pandas_drop]

### 三-将数据切分为训练数据和测试数据 ###

from sklearn.cross_validation import train_test_split
    Xtrain, Xtest, ytrain, ytest = train_test_split(X_iris, y_iris, random_state=1)

`cross validation`是**交叉验证**的意思, [参考文章][Link 1]  
参数`random_state`是固定的随机种子, [参考文章][Link 2]

### 四-调用高斯朴素贝叶斯实现训练 ###

from sklearn.naive_bayes import GaussianNB
    model = GaussianNB()
    model.fit(Xtrain, ytrain)
    y_model = model.predict(Xtest) # 进行预测

调用`Gaussian naive Bayes`模型, 并进行拟合, 预测

### 五-对测试结果进行评估 ###

[评估方法介绍][Link 3]

from sklearn.metrics import accuracy_score
    accuracy_score(ytest, y_model)

得到准确率为`0.9736842105263158`  
`metrics`是指标的意思.

说明对特征明显的数据, 即使是非常简单的分类算法也可以高效地进行分析.

[github]: https://github.com/mwaskom/seaborn-data
[pandas_drop]: https://blog.csdn.net/sinat_32547403/article/details/73822660
[Link 1]: https://blog.csdn.net/CherDW/article/details/54881167
[Link 2]: http://sofasofa.io/forum_main_post.php?postid=1001204
[Link 3]: https://blog.csdn.net/CherDW/article/details/55813071