如何将数据分成 3 组(训练、验证和测试)? [英] How to split data into 3 sets (train, validation and test)?

问题描述

我有一个 Pandas 数据框，我希望将它分成 3 个独立的集合.我知道使用 sklearn 中的 train_test_split.cross_validation，可以将数据分成两组(训练和测试).但是，我找不到将数据分成三组的任何解决方案.最好，我想要原始数据的索引.

I have a pandas dataframe and I wish to divide it to 3 separate sets. I know that using train_test_split from sklearn.cross_validation, one can divide the data in two sets (train and test). However, I couldn't find any solution about splitting the data into three sets. Preferably, I'd like to have the indices of the original data.

我知道一种解决方法是使用 train_test_split 两次并以某种方式调整索引.但是是否有更标准/内置的方法将数据拆分为 3 组而不是 2 组?

I know that a workaround would be to use train_test_split two times and somehow adjust the indices. But is there a more standard / built-in way to split the data into 3 sets instead of 2?

推荐答案

Numpy 解决方案.我们将首先打乱整个数据集(df.sample(frac=1, random_state=42))，然后将我们的数据集拆分为以下部分:

Numpy solution. We will shuffle the whole dataset first (df.sample(frac=1, random_state=42)) and then split our data set into the following parts:

• 60% - 训练集，
• 20% - 验证集，
• 20% - 测试集

In [305]: train, validate, test = 
              np.split(df.sample(frac=1, random_state=42), 
                       [int(.6*len(df)), int(.8*len(df))])

In [306]: train
Out[306]:
          A         B         C         D         E
0  0.046919  0.792216  0.206294  0.440346  0.038960
2  0.301010  0.625697  0.604724  0.936968  0.870064
1  0.642237  0.690403  0.813658  0.525379  0.396053
9  0.488484  0.389640  0.599637  0.122919  0.106505
8  0.842717  0.793315  0.554084  0.100361  0.367465
7  0.185214  0.603661  0.217677  0.281780  0.938540

In [307]: validate
Out[307]:
          A         B         C         D         E
5  0.806176  0.008896  0.362878  0.058903  0.026328
6  0.145777  0.485765  0.589272  0.806329  0.703479

In [308]: test
Out[308]:
          A         B         C         D         E
4  0.521640  0.332210  0.370177  0.859169  0.401087
3  0.333348  0.964011  0.083498  0.670386  0.169619

[int(.6*len(df)), int(.8*len(df))] - 是 indices_or_sections 数组"http://docs.scipy.org/doc/numpy/reference/generated/numpy.split.html" rel="noreferrer">numpy.split().

[int(.6*len(df)), int(.8*len(df))] - is an indices_or_sections array for numpy.split().

这里是 np.split() 用法的一个小演示 - 让我们将 20 个元素的数组分成以下部分:80%、10%、10%:

Here is a small demo for np.split() usage - let's split 20-elements array into the following parts: 80%, 10%, 10%:

In [45]: a = np.arange(1, 21)

In [46]: a
Out[46]: array([ 1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20])

In [47]: np.split(a, [int(.8 * len(a)), int(.9 * len(a))])
Out[47]:
[array([ 1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16]),
 array([17, 18]),
 array([19, 20])]

这篇关于如何将数据分成 3 组(训练、验证和测试)?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！