Can dictionary data split into test and training set randomly?

Question

I want to understand if I have a set of Dictionary data in JSON such as example below:

data = {'a':'120120121',
'b':'12301101',
'c':'120120121',
'd':'12301101',
'e':'120120121',
'f':'12301101',
'g':'120120121',
'h':'12301101',
'i':'120120121',
'j':'12301101'}

Is it possible to split the dictionary to 70:30 randomly using Python?

The output should be like:

training_data = {'a':'120120121',
    'b':'12301101',
    'c':'120120121',
    'e':'120120121',
    'g':'120120121',
    'i':'120120121',
    'j':'12301101'}

test_data = {'d':'12301101',
    'f':'12301101',
    'h':'12301101'}

Answer 1

The easiest way would be to just use sklearn.model_selection.train_test_split here, and turn back to dictionary if that is the structure you want:

from sklearn.model_selection import train_test_split

s = pd.Series(data)
training_data , test_data  = [i.to_dict() for i in train_test_split(s, train_size=0.7)]

---

print(training_data)
# {'b': '12301101', 'j': '12301101', 'a': '120120121', 'f': '12301101', 
# 'e': '120120121', 'c': '120120121', 'h': '12301101'}

print(test_data)
# {'i': '120120121', 'd': '12301101', 'g': '120120121'}