How to split data into 3 sets (train, validation and test)?

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% - train set,
• 20% - validation set,
• 20% - test set

In [305]: train, validate, test = \              np.split(df.sample(frac=1, random_state=42),                        [int(.6*len(df)), int(.8*len(df))])In [306]: trainOut[306]:          A         B         C         D         E0  0.046919  0.792216  0.206294  0.440346  0.0389602  0.301010  0.625697  0.604724  0.936968  0.8700641  0.642237  0.690403  0.813658  0.525379  0.3960539  0.488484  0.389640  0.599637  0.122919  0.1065058  0.842717  0.793315  0.554084  0.100361  0.3674657  0.185214  0.603661  0.217677  0.281780  0.938540In [307]: validateOut[307]:          A         B         C         D         E5  0.806176  0.008896  0.362878  0.058903  0.0263286  0.145777  0.485765  0.589272  0.806329  0.703479In [308]: testOut[308]:          A         B         C         D         E4  0.521640  0.332210  0.370177  0.859169  0.4010873  0.333348  0.964011  0.083498  0.670386  0.169619

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

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]: aOut[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])]

Note:

Function was written to handle seeding of randomized set creation. You should not rely on set splitting that doesn't randomize the sets.

import numpy as npimport pandas as pddef 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

Demonstration

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

However, one approach to dividing the dataset into train, test, cv with 0.6, 0.2, 0.2 would be to use the train_test_split method twice.

from sklearn.model_selection import train_test_splitx, 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)