注意:

函数被编写来处理随机集创建的播种。你不应该依赖集分割，它不会随机化集合。

import numpy as np
import pandas as pd

def train_validate_test_split(df, train_percent=.6, validate_percent=.2, seed=None):
    np.random.seed(seed)
    perm = np.random.permutation(df.index)
    m = len(df.index)
    train_end = int(train_percent * m)
    validate_end = int(validate_percent * m) + train_end
    train = df.iloc[perm[:train_end]]
    validate = df.iloc[perm[train_end:validate_end]]
    test = df.iloc[perm[validate_end:]]
    return train, validate, test

示范

np.random.seed([3,1415])
df = pd.DataFrame(np.random.rand(10, 5), columns=list('ABCDE'))
df

train, validate, test = train_validate_test_split(df)

train

validate

test

考虑到df id你的原始数据帧:

1 -首先你在训练和测试之间分割数据(10%):

my_test_size = 0.10

X_train_, X_test, y_train_, y_test = train_test_split(
    df.index.values,
    df.label.values,
    test_size=my_test_size,
    random_state=42,
    stratify=df.label.values,    
)

2 -然后你在训练和验证之间分割训练集(20%):

my_val_size = 0.20

X_train, X_val, y_train, y_val = train_test_split(
    df.loc[X_train_].index.values,
    df.loc[X_train_].label.values,
    test_size=my_val_size,
    random_state=42,
    stratify=df.loc[X_train_].label.values,  
)

3 -然后，根据上述步骤中生成的索引对原始数据帧进行切片:

# data_type is not necessary. 
df['data_type'] = ['not_set']*df.shape[0]
df.loc[X_train, 'data_type'] = 'train'
df.loc[X_val, 'data_type'] = 'val'
df.loc[X_test, 'data_type'] = 'test'

结果是这样的:

注意:此解决方案使用问题中提到的解决方案。

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])]

将数据集分割为训练集和测试集，如在其他答案中一样，使用

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

然后，如果您适合您的模型，您可以添加validation_split作为参数。这样就不需要提前创建验证集。例如:

from tensorflow.keras import Model

model = Model(input_layer, out)

[...]

history = model.fit(x=X_train, y=y_train, [...], validation_split = 0.3)

验证集旨在作为训练集训练期间的代表运行测试集，完全来自训练集，无论是通过k-fold交叉验证(推荐)还是通过validation_split;然后，您不需要单独创建一个验证集，仍然可以将数据集分为您所要求的三个集。

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]。