Scikit 学习多标签分类，从 MultiLabelBinarizer 中取回标签 [英] Scikit Learn Multilabel Classification, Getting back labels from MultiLabelBinarizer

问题描述

在多标签分类问题中，我使用 MultiLabelBinarizer 将我的 20 个文本标签转换为一个由 0 和 1 组成的二进制列表.

In a multilabel classification problem, i use MultiLabelBinarizer to transform my 20 text labels into a binary list of zeros and ones.

预测后，我得到了 20 个二进制值的列表，我想输出相应的文本标签.

After prediction I get my list of 20 binary values, and I would like to output the corresponding text labels.

我只是想知道 MultiLabelBinarizer() 是否提供了返回转换，或者我应该手动进行.

I am just wondering whether MultiLabelBinarizer() provides a getting back transformation or I should do it manually.

推荐答案

是的，MultiLabelBinarizer 提供了一个名为 inverse_transform() 的方法，它将二值化标签转换回提供给的原始名称它在 fit() 期间.

Yes, The MultiLabelBinarizer provides a method named inverse_transform() which will convert the binarized labels back to the original names that were supplied to it during fit().

http://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MultiLabelBinarizer.html#sklearn.preprocessing.MultiLabelBinarizer.inverse_transform

inverse_transform(yt)

inverse_transform(yt)

Transform the given indicator matrix into label sets
Parameters:   

yt : array or sparse matrix of shape (n_samples, n_classes)

    A matrix containing only 1s ands 0s.

Returns:  

y : list of tuples

    The set of labels for each sample such that y[i] consists of classes_[j] for each yt[i, j] == 1.

这篇关于Scikit 学习多标签分类，从 MultiLabelBinarizer 中取回标签的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！