问题描述
我正在尝试使用UCI存储库中的汽车评估数据集，我想知道是否存在方便的方法来对sklearn中的分类变量进行二值化.一种方法是使用LabelBinarizer的DictVectorizer，但在这里我得到了k个不同的特征，而为了避免共线性化，您应该只有k-1个. 我想我可以编写自己的函数并删除一列，但这种簿记工作很繁琐，是否有一种简单的方法来执行此类转换并因此获得稀疏矩阵?
I'm trying to use the car evaluation dataset from the UCI repository and I wonder whether there is a convenient way to binarize categorical variables in sklearn. One approach would be to use the DictVectorizer of LabelBinarizer but here I'm getting k different features whereas you should have just k-1 in order to avoid collinearization. I guess I could write my own function and drop one column but this bookkeeping is tedious, is there an easy way to perform such transformations and get as a result a sparse matrix?
推荐答案
DictVectorizer是生成分类变量的一键编码的推荐方法.您可以使用sparse参数创建一个稀疏的CSR矩阵，而不是一个密集的numpy数组.我通常不关心多重共线性，也没有注意到我倾向于使用的方法(即LinearSVC，SGDClassifier，基于树的方法)存在问题.
DictVectorizer is the recommended way to generate a one-hot encoding of categorical variables; you can use the sparse argument to create a sparse CSR matrix instead of a dense numpy array. I usually don't care about multicollinearity and I haven't noticed a problem with the approaches that I tend to use (i.e. LinearSVC, SGDClassifier, Tree-based methods).
打补丁DictVectorizer以便为每个分类功能删除一列不是问题-您只需在fit方法的末尾从DictVectorizer.vocabulary中删除一个术语. (随时欢迎拉动请求！)
It shouldn't be a problem to patch the DictVectorizer to drop one column per categorical feature - you simple need to remove one term from DictVectorizer.vocabulary at the end of the fit method. (Pull requests are always welcome!)
这篇关于如何在sklearn中编码分类变量?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！