下面是一个Python函数，它将Pandas数据帧划分为分层抽样的训练、验证和测试数据帧。它通过两次调用scikit-learn的函数train_test_split()来执行这种分割。

import pandas as pd
from sklearn.model_selection import train_test_split

def split_stratified_into_train_val_test(df_input, stratify_colname='y',
                                         frac_train=0.6, frac_val=0.15, frac_test=0.25,
                                         random_state=None):
    '''
    Splits a Pandas dataframe into three subsets (train, val, and test)
    following fractional ratios provided by the user, where each subset is
    stratified by the values in a specific column (that is, each subset has
    the same relative frequency of the values in the column). It performs this
    splitting by running train_test_split() twice.

    Parameters
    ----------
    df_input : Pandas dataframe
        Input dataframe to be split.
    stratify_colname : str
        The name of the column that will be used for stratification. Usually
        this column would be for the label.
    frac_train : float
    frac_val   : float
    frac_test  : float
        The ratios with which the dataframe will be split into train, val, and
        test data. The values should be expressed as float fractions and should
        sum to 1.0.
    random_state : int, None, or RandomStateInstance
        Value to be passed to train_test_split().

    Returns
    -------
    df_train, df_val, df_test :
        Dataframes containing the three splits.
    '''

    if frac_train + frac_val + frac_test != 1.0:
        raise ValueError('fractions %f, %f, %f do not add up to 1.0' % \
                         (frac_train, frac_val, frac_test))

    if stratify_colname not in df_input.columns:
        raise ValueError('%s is not a column in the dataframe' % (stratify_colname))

    X = df_input # Contains all columns.
    y = df_input[[stratify_colname]] # Dataframe of just the column on which to stratify.

    # Split original dataframe into train and temp dataframes.
    df_train, df_temp, y_train, y_temp = train_test_split(X,
                                                          y,
                                                          stratify=y,
                                                          test_size=(1.0 - frac_train),
                                                          random_state=random_state)

    # Split the temp dataframe into val and test dataframes.
    relative_frac_test = frac_test / (frac_val + frac_test)
    df_val, df_test, y_val, y_test = train_test_split(df_temp,
                                                      y_temp,
                                                      stratify=y_temp,
                                                      test_size=relative_frac_test,
                                                      random_state=random_state)

    assert len(df_input) == len(df_train) + len(df_val) + len(df_test)

    return df_train, df_val, df_test

下面是一个完整的工作示例。

考虑一个数据集，该数据集具有一个标签，您希望在其上执行分层。这个标签在原始数据集中有自己的分布，比如75% foo, 15% bar和10% baz。现在，让我们使用60/20/20的比例将数据集分割为训练、验证和测试子集，其中每个分割保留相同的标签分布。如下图所示:

下面是示例数据集:

df = pd.DataFrame( { 'A': list(range(0, 100)),
                     'B': list(range(100, 0, -1)),
                     'label': ['foo'] * 75 + ['bar'] * 15 + ['baz'] * 10 } )

df.head()
#    A    B label
# 0  0  100   foo
# 1  1   99   foo
# 2  2   98   foo
# 3  3   97   foo
# 4  4   96   foo

df.shape
# (100, 3)

df.label.value_counts()
# foo    75
# bar    15
# baz    10
# Name: label, dtype: int64

现在，让我们从上面调用split_stratified_into_train_val_test()函数，按照60/20/20的比例获取训练、验证和测试数据帧。

df_train, df_val, df_test = \
    split_stratified_into_train_val_test(df, stratify_colname='label', frac_train=0.60, frac_val=0.20, frac_test=0.20)

三个数据帧df_train、df_val和df_test包含所有原始行，但它们的大小将遵循上面的比例。

df_train.shape
#(60, 3)

df_val.shape
#(20, 3)

df_test.shape
#(20, 3)

此外，三次分割中的每一次都将具有相同的标签分布，即75% foo, 15% bar和10% baz。

df_train.label.value_counts()
# foo    45
# bar     9
# baz     6
# Name: label, dtype: int64

df_val.label.value_counts()
# foo    15
# bar     3
# baz     2
# Name: label, dtype: int64

df_test.label.value_counts()
# foo    15
# bar     3
# baz     2
# Name: label, dtype: int64

使用train_test_split非常方便，不需要在划分到几个集后执行重新索引，也不需要编写一些额外的代码。上面的最佳答案没有提到使用train_test_split分隔两次而不改变分区大小将不会给出最初预期的分区:

x_train, x_remain = train_test_split(x, test_size=(val_size + test_size))

那么x_remain中的验证集和测试集的部分就会发生变化，可以算作

new_test_size = np.around(test_size / (val_size + test_size), 2)
# To preserve (new_test_size + new_val_size) = 1.0 
new_val_size = 1.0 - new_test_size

x_val, x_test = train_test_split(x_remain, test_size=new_test_size)

在这种情况下，将保存所有初始分区。

然而，将数据集分为train、test、cv(0.6、0.2、0.2)的一种方法是使用train_test_split方法两次。

from sklearn.model_selection import train_test_split

x, x_test, y, y_test = train_test_split(xtrain,labels,test_size=0.2,train_size=0.8)
x_train, x_cv, y_train, y_cv = train_test_split(x,y,test_size = 0.25,train_size =0.75)

我能想到的最简单的方法是将分割分数映射到数组下标，如下所示:

train_set = data[:int((len(data)+1)*train_fraction)]
test_set = data[int((len(data)+1)*train_fraction):int((len(data)+1)*(train_fraction+test_fraction))]
val_set = data[int((len(data)+1)*(train_fraction+test_fraction)):]

其中data = random.shuffle(data)

回答任意数量的子集:

def _separate_dataset(patches, label_patches, percentage, shuffle: bool = True):
    """
    :param patches: data patches
    :param label_patches: label patches
    :param percentage: list of percentages for each value, example [0.9, 0.02, 0.08] to get 90% train, 2% val and 8% test.
    :param shuffle: Shuffle dataset before split.
    :return: tuple of two lists of size = len(percentage), one with data x and other with labels y.
    """
    x_test = patches
    y_test = label_patches
    percentage = list(percentage)       # need it to be mutable
    assert sum(percentage) == 1., f"percentage must add to 1, but it adds to sum{percentage} = {sum(percentage)}"
    x = []
    y = []
    for i, per in enumerate(percentage[:-1]):
        x_train, x_test, y_train, y_test = train_test_split(x_test, y_test, test_size=1-per, shuffle=shuffle)
        percentage[i+1:] = [value / (1-percentage[i]) for value in percentage[i+1:]]
        x.append(x_train)
        y.append(y_train)
    x.append(x_test)
    y.append(y_test)
    return x, y

这适用于任何比例。在本例中，您应该执行percentage = [train_percentage, val_percentage, test_percentage]。

Numpy解决方案。我们将首先洗牌整个数据集(df。Sample (frac=1, random_state=42))，然后将我们的数据集分成以下部分:

60% -列车集， 20% -验证集， 20% -测试装置

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numpy.split()的indices_or_sections数组。

下面是一个np.split()使用的小演示-让我们把20个元素的数组分成以下部分: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])]