pandas 将分类变量从 String 转换为 int 表示
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/41078003/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Convert categorical variables from String to int representation
提问by Abhi
I have a numpy array of classification of text in the form of String array, i.e.
y_train = ['A', 'B', 'A', 'C',...]. I am trying to apply SKlearn multinomial NB algorithm to predict classes for entire dataset.
我有一个字符串数组形式的文本分类的 numpy 数组，即
y_train = ['A', 'B', 'A', 'C',...]. 我正在尝试应用 SKlearn 多项 NB 算法来预测整个数据集的类别。
I want to convert the String classes into integers to be able to input into the algorithm and convert ['A', 'B', 'A', 'C', ...]into ['1', '2', '1', '3', ...]
我想将字符串类转换为整数，以便能够输入算法并转换['A', 'B', 'A', 'C', ...]为['1', '2', '1', '3', ...]
I can write a for loop to go through array and create a new one with int classifiers but is there a direct function to achieve this
我可以编写一个 for 循环来遍历数组并使用 int 分类器创建一个新的循环，但是是否有直接的函数来实现这一点
采纳答案by Ted Petrou
If you are using sklearn, I would suggest sticking with methods in that library that do these things for you. Sklearn has a number of ways of preprocessing data such as encoding labels. One of which is the sklearn.preprocessing.LabelEncoderfunction.
如果您正在使用 sklearn，我建议您坚持使用该库中为您执行这些操作的方法。Sklearn 有多种预处理数据的方法，例如编码标签。其中之一是sklearn.preprocessing.LabelEncoder功能。
from sklearn.preprocessing import LabelEncoder
le = LabelEncoder()
le.fit_transform(y_train)
Which outputs
哪些输出
array([0, 1, 0, 2])
Use le.inverse_transform([0,1,2])to map back
使用le.inverse_transform([0,1,2])映射回
回答by MaxU
Try factorizemethod:
尝试分解方法：
In [264]: y_train = pd.Series(['A', 'B', 'A', 'C'])
In [265]: y_train
Out[265]:
0 A
1 B
2 A
3 C
dtype: object
In [266]: pd.factorize(y_train)
Out[266]: (array([0, 1, 0, 2], dtype=int64), Index(['A', 'B', 'C'], dtype='object'))
Demo:
演示：
In [271]: fct = pd.factorize(y_train)[0]+1
In [272]: fct
Out[272]: array([1, 2, 1, 3], dtype=int64)
