Python之sklearn：LabelEncoder函数简介(编码与编码还原)、使用方法、具体案例之详细攻略

电玩女神 2022-11-21 11:22 400阅读 0赞

**目录**

LabelEncoder函数的简介(编码与编码还原)

Methods

LabelEncoder函数的使用方法

LabelEncoder函数的具体案例

1、基础案例

2、在数据缺失和test数据内存在新值(train数据未出现过)环境下的数据LabelEncoder化

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

## LabelEncoder函数的简介(编码与编码还原) ##

<table> 
 <tbody> 
 <tr> 
 <td style="width:470px;"> class LabelEncoder Found at: sklearn.preprocessing._labelclass LabelEncoder(TransformerMixin, BaseEstimator): &nbsp; &nbsp; """Encode target labels with value between 0 and n_classes-1. &nbsp; &nbsp; This transformer should be used to encode target values, *i.e.* `y`, and not the input `X`. &nbsp; &nbsp; Read more in the :ref:`User Guide &lt;preprocessing_targets&gt;`. </td> 
 <td style="width:428px;"> &nbsp; ""对目标标签进行编码，值在0到n_class -1之间。 这个转换器应该用于编码目标值，*即' y '，而不是输入' X '。 更多内容见:ref: ' User Guide '。 </td> 
 </tr> 
 <tr> 
 <td style="width:470px;">&nbsp; &nbsp; .. versionadded:: 0.12 &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; Attributes &nbsp; &nbsp; ---------- &nbsp; &nbsp; classes_ : array of shape (n_class,) &nbsp; &nbsp; Holds the label for each class. &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; Examples &nbsp; &nbsp; -------- &nbsp; &nbsp; `LabelEncoder` can be used to normalize labels. &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; &gt;&gt;&gt; from sklearn import preprocessing &nbsp; &nbsp; &gt;&gt;&gt; le = preprocessing.LabelEncoder() &nbsp; &nbsp; &gt;&gt;&gt; le.fit([1, 2, 2, 6]) &nbsp; &nbsp; LabelEncoder() &nbsp; &nbsp; &gt;&gt;&gt; le.classes_ &nbsp; &nbsp; array([1, 2, 6]) &nbsp; &nbsp; &gt;&gt;&gt; le.transform([1, 1, 2, 6]) &nbsp; &nbsp; array([0, 0, 1, 2]...) &nbsp; &nbsp; &gt;&gt;&gt; le.inverse_transform([0, 0, 1, 2]) &nbsp; &nbsp; array([1, 1, 2, 6]) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; It can also be used to transform non-numerical labels (as long as they are hashable and comparable) to numerical labels. &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; &gt;&gt;&gt; le = preprocessing.LabelEncoder() &nbsp; &nbsp; &gt;&gt;&gt; le.fit(["paris", "paris", "tokyo", "amsterdam"]) &nbsp; &nbsp; LabelEncoder() &nbsp; &nbsp; &gt;&gt;&gt; list(le.classes_) &nbsp; &nbsp; ['amsterdam', 'paris', 'tokyo'] &nbsp; &nbsp; &gt;&gt;&gt; le.transform(["tokyo", "tokyo", "paris"]) &nbsp; &nbsp; array([2, 2, 1]...) &nbsp; &nbsp; &gt;&gt;&gt; list(le.inverse_transform([2, 2, 1])) &nbsp; &nbsp; ['tokyo', 'tokyo', 'paris'] &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; See also &nbsp; &nbsp; -------- &nbsp; &nbsp; sklearn.preprocessing.OrdinalEncoder : Encode categorical features using an ordinal encoding scheme. &nbsp; &nbsp; sklearn.preprocessing.OneHotEncoder : Encode categorical features as a one-hot numeric array.</td> 
 <td style="width:428px;"> . .versionadded:: 0.12 属性 ---------- classes_:形状数组(n_class，) 保存每个类的标签。 例子 -------- “LabelEncoder”可用于规范化标签。 &nbsp; &nbsp; &nbsp; &gt;&gt;&gt; from sklearn import preprocessing &nbsp; &nbsp; &gt;&gt;&gt; le = preprocessing.LabelEncoder() &nbsp; &nbsp; &gt;&gt;&gt; le.fit([1, 2, 2, 6]) &nbsp; &nbsp; LabelEncoder() &nbsp; &nbsp; &gt;&gt;&gt; le.classes_ &nbsp; &nbsp; array([1, 2, 6]) &nbsp; &nbsp; &gt;&gt;&gt; le.transform([1, 1, 2, 6]) &nbsp; &nbsp; array([0, 0, 1, 2]...) &nbsp; &nbsp; &gt;&gt;&gt; le.inverse_transform([0, 0, 1, 2]) &nbsp; &nbsp; array([1, 1, 2, 6]) 它还可以用于将非数字标签(只要它们是可hashable和可比的)转换为数字标签。 &nbsp; &nbsp; &nbsp; &gt;&gt;&gt; le = preprocessing.LabelEncoder() &nbsp; &nbsp; &gt;&gt;&gt; le.fit(["paris", "paris", "tokyo", "amsterdam"]) &nbsp; &nbsp; LabelEncoder() &nbsp; &nbsp; &gt;&gt;&gt; list(le.classes_) &nbsp; &nbsp; ['amsterdam', 'paris', 'tokyo'] &nbsp; &nbsp; &gt;&gt;&gt; le.transform(["tokyo", "tokyo", "paris"]) &nbsp; &nbsp; array([2, 2, 1]...) &nbsp; &nbsp; &gt;&gt;&gt; list(le.inverse_transform([2, 2, 1])) &nbsp; &nbsp; ['tokyo', 'tokyo', 'paris'] 另请参阅 -------- sklearn.preprocessing.OrdinalEncoder :序号编码器:使用序号编码方案编码分类特征。 sklearn.preprocessing.OneHotEncoder :&nbsp;&nbsp;将分类特性编码为一个热的数字数组。 </td> 
 </tr> 
 <tr> 
 <td style="width:470px;"> &nbsp; &nbsp; """ &nbsp; &nbsp; def fit(self, y): &nbsp; &nbsp; &nbsp; &nbsp; """Fit label encoder &nbsp; &nbsp; &nbsp; &nbsp; Parameters &nbsp; &nbsp; &nbsp; &nbsp; ---------- &nbsp; &nbsp; &nbsp; &nbsp; y : array-like of shape (n_samples,) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Target values. &nbsp; &nbsp; &nbsp; &nbsp; Returns &nbsp; &nbsp; &nbsp; &nbsp; ------- &nbsp; &nbsp; &nbsp; &nbsp; self : returns an instance of self. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; y = column_or_1d(y, warn=True) &nbsp; &nbsp; &nbsp; &nbsp; self.classes_ = _encode(y) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def fit_transform(self, y): &nbsp; &nbsp; &nbsp; &nbsp; """Fit label encoder and return encoded labels &nbsp; &nbsp; &nbsp; &nbsp; Parameters &nbsp; &nbsp; &nbsp; &nbsp; ---------- &nbsp; &nbsp; &nbsp; &nbsp; y : array-like of shape [n_samples] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Target values. &nbsp; &nbsp; &nbsp; &nbsp; Returns &nbsp; &nbsp; &nbsp; &nbsp; ------- &nbsp; &nbsp; &nbsp; &nbsp; y : array-like of shape [n_samples] &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; y = column_or_1d(y, warn=True) &nbsp; &nbsp; &nbsp; &nbsp; self.classes_, y = _encode(y, encode=True) &nbsp; &nbsp; &nbsp; &nbsp; return y &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def transform(self, y): &nbsp; &nbsp; &nbsp; &nbsp; """Transform labels to normalized encoding. &nbsp; &nbsp; &nbsp; &nbsp; Parameters &nbsp; &nbsp; &nbsp; &nbsp; ---------- &nbsp; &nbsp; &nbsp; &nbsp; y : array-like of shape [n_samples] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Target values. &nbsp; &nbsp; &nbsp; &nbsp; Returns &nbsp; &nbsp; &nbsp; &nbsp; ------- &nbsp; &nbsp; &nbsp; &nbsp; y : array-like of shape [n_samples] &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; check_is_fitted(self) &nbsp; &nbsp; &nbsp; &nbsp; y = column_or_1d(y, warn=True) &nbsp; &nbsp; &nbsp; &nbsp; # transform of empty array is empty array &nbsp; &nbsp; &nbsp; &nbsp; if _num_samples(y) == 0: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; return np.array([]) &nbsp; &nbsp; &nbsp; &nbsp; _, y = _encode(y, uniques=self.classes_, encode=True) &nbsp; &nbsp; &nbsp; &nbsp; return y &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def inverse_transform(self, y): &nbsp; &nbsp; &nbsp; &nbsp; """Transform labels back to original encoding. &nbsp; &nbsp; &nbsp; &nbsp; Parameters &nbsp; &nbsp; &nbsp; &nbsp; ---------- &nbsp; &nbsp; &nbsp; &nbsp; y : numpy array of shape [n_samples] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Target values. &nbsp; &nbsp; &nbsp; &nbsp; Returns &nbsp; &nbsp; &nbsp; &nbsp; ------- &nbsp; &nbsp; &nbsp; &nbsp; y : numpy array of shape [n_samples] &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; check_is_fitted(self) &nbsp; &nbsp; &nbsp; &nbsp; y = column_or_1d(y, warn=True) &nbsp; &nbsp; &nbsp; &nbsp; # inverse transform of empty array is empty array &nbsp; &nbsp; &nbsp; &nbsp; if _num_samples(y) == 0: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; return np.array([]) &nbsp; &nbsp; &nbsp; &nbsp; diff = np.setdiff1d(y, np.arange(len(self.classes_))) &nbsp; &nbsp; &nbsp; &nbsp; if len(diff): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise ValueError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "y contains previously unseen labels: %s" % str(diff)) &nbsp; &nbsp; &nbsp; &nbsp; y = np.asarray(y) &nbsp; &nbsp; &nbsp; &nbsp; return self.classes_[y] &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _more_tags(self): &nbsp; &nbsp; &nbsp; &nbsp; return {'X_types':['1dlabels']} </td> 
 <td style="width:428px;">&nbsp;</td> 
 </tr> 
 </tbody> 
</table>

### Methods ###

<table> 
 <tbody> 
 <tr> 
 <td> <a href="https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html?highlight=labelencoder#sklearn.preprocessing.LabelEncoder.fit" rel="nofollow"><code>fit</code></a>(y) </td> 
 <td> Fit label encoder </td> 
 </tr> 
 <tr> 
 <td> <a href="https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html?highlight=labelencoder#sklearn.preprocessing.LabelEncoder.fit_transform" rel="nofollow"><code>fit_transform</code></a>(y) </td> 
 <td> Fit label encoder and return encoded labels </td> 
 </tr> 
 <tr> 
 <td> <a href="https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html?highlight=labelencoder#sklearn.preprocessing.LabelEncoder.get_params" rel="nofollow"><code>get_params</code></a>([deep]) </td> 
 <td> Get parameters for this estimator. </td> 
 </tr> 
 <tr> 
 <td> <a href="https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html?highlight=labelencoder#sklearn.preprocessing.LabelEncoder.inverse_transform" rel="nofollow"><code>inverse_transform</code></a>(y) </td> 
 <td> Transform labels back to original encoding. </td> 
 </tr> 
 <tr> 
 <td> <a href="https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html?highlight=labelencoder#sklearn.preprocessing.LabelEncoder.set_params" rel="nofollow"><code>set_params</code></a>(**params) </td> 
 <td> Set the parameters of this estimator. </td> 
 </tr> 
 <tr> 
 <td> <a href="https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html?highlight=labelencoder#sklearn.preprocessing.LabelEncoder.transform" rel="nofollow"><code>transform</code></a>(y) </td> 
 <td> Transform labels to normalized encoding. </td> 
 </tr> 
 </tbody> 
</table>

## LabelEncoder函数的使用方法 ##

import pandas as pd
    from sklearn.preprocessing import LabelEncoder
    from DataScienceNYY.DataAnalysis import dataframe_fillAnyNull,Dataframe2LabelEncoder
    
    
    #构造数据
    train_data_dict={'Name':['张三','李四','王五','赵六','张七','李八','王十','un'],
                    'Age':[22,23,24,25,22,22,22,None],
                    'District':['北京','上海','广东','深圳','山东','河南','浙江',' '],
                    'Job':['CEO','CTO','CFO','COO','CEO','CTO','CEO','']}
    test_data_dict={'Name':['张三','李四','王十一',None],
                    'Age':[22,23,22,'un'],
                    'District':['北京','上海','广东',''],
                    'Job':['CEO','CTO','UFO',' ']}
    train_data_df = pd.DataFrame(train_data_dict)
    test_data_df = pd.DataFrame(test_data_dict)
    print(train_data_df,'\n',test_data_df)
    
    
    #缺失数据填充
    for col in train_data_df.columns:
            train_data_df[col]=dataframe_fillAnyNull(train_data_df,col)
            test_data_df[col]=dataframe_fillAnyNull(test_data_df,col)
    print(train_data_df,'\n',test_data_df)
    
    
    #数据LabelEncoder化
    train_data,test_data=Dataframe2LabelEncoder(train_data_df,test_data_df)
    print(train_data,'\n',test_data)

## LabelEncoder函数的具体案例 ##

### 1、基础案例 ###

LabelEncoder can be used to normalize labels.
    
    >>>
    >>> from sklearn import preprocessing
    >>> le = preprocessing.LabelEncoder()
    >>> le.fit([1, 2, 2, 6])
    LabelEncoder()
    >>> le.classes_
    array([1, 2, 6])
    >>> le.transform([1, 1, 2, 6])
    array([0, 0, 1, 2]...)
    >>> le.inverse_transform([0, 0, 1, 2])
    array([1, 1, 2, 6])
    It can also be used to transform non-numerical labels (as long as they are hashable and comparable) to numerical labels.
    
    >>>
    >>> le = preprocessing.LabelEncoder()
    >>> le.fit(["paris", "paris", "tokyo", "amsterdam"])
    LabelEncoder()
    >>> list(le.classes_)
    ['amsterdam', 'paris', 'tokyo']
    >>> le.transform(["tokyo", "tokyo", "paris"])
    array([2, 2, 1]...)
    >>> list(le.inverse_transform([2, 2, 1]))
    ['tokyo', 'tokyo', 'paris']

### 2、在数据缺失和test数据内存在新值(train数据未出现过)环境下的数据LabelEncoder化 ###

**参考文章**：[Python之sklearn：LabelEncoder函数的使用方法之使用LabelEncoder之前的必要操作][Python_sklearn_LabelEncoder_LabelEncoder]

import numpy as np
    from sklearn.preprocessing import LabelEncoder
    
    #训练train数据
    LE= LabelEncoder()
    LE.fit(train_df[col])
    
    #test数据中的新值添加到LE.classes_
    test_df[col] =test_df[col].map(lambda s:'Unknown' if s not in LE.classes_ else s) 
    LE.classes_ = np.append(LE.classes_, 'Unknown') 
     
    #分别转化train、test数据
    train_df[col] = LE.transform(train_df[col]) 
    test_df[col] = LE.transform(test_df[col])

[Python_sklearn_LabelEncoder_LabelEncoder]: https://yunyaniu.blog.csdn.net/article/details/109408854