XGBoost:将 dmatrix 转换为 numpy.array

问题描述

我想检查 DMatrix 对象.文档提到获取行数的方法和列以及每行的切片:

I'd like to inspect the DMatrix object. The documentation mentions the methods to get the number of rows and columns and also the slice at each row:

dmatrix.slice(index)

但我想使用 numpy 进行更复杂的转换.有没有一种简单的方法可以将 DMatrix 转换为 numpy 数组?

But I'd like to do more complex transformations using numpy. Is there a simple way to convert the DMatrix to a numpy array?

推荐答案

DMatrix 是一条单行道！获得 DMatrix 后，您无法将其内容作为 numpy.ndarray 或 pandas.DataFrame 取回.

DMatrix is a one-way street! Once you get DMatrix, you cannot get back its content as numpy.ndarray or pandas.DataFrame.

[更新] 一个库仅用于将 XGBoost 的 DMatrix 格式转换为 numpy.array 的特定任务:https://github.com/aporia-ai/dmatrix2np

[Update] A library is implemented only for the particular task of convert XGBoost's DMatrix format to numpy.array: https://github.com/aporia-ai/dmatrix2np

这篇关于XGBoost:将 dmatrix 转换为 numpy.array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！