i faced this problem while i was strat sampling the dataset-CodePudding

In [16] : Strat_d3=d3.groupby('Label', group_keys=False).apply(lambda x: x.sample(1000))
Traceback (most recent call last):

  File "<ipython-input-16-f54910ba8f95>", line 1, in <module>
    Strat_d3=d3.groupby('Label', group_keys=False).apply(lambda x: x.sample(1000))

  File "C:\Users\Msi\anaconda3\lib\site-packages\pandas\core\groupby\groupby.py", line 894, in apply
    result = self._python_apply_general(f, self._selected_obj)

  File "C:\Users\Msi\anaconda3\lib\site-packages\pandas\core\groupby\groupby.py", line 928, in _python_apply_general
    keys, values, mutated = self.grouper.apply(f, data, self.axis)

  File "C:\Users\Msi\anaconda3\lib\site-packages\pandas\core\groupby\ops.py", line 238, in apply
    res = f(group)

  File "<ipython-input-16-f54910ba8f95>", line 1, in <lambda>
    Strat_d3=d3.groupby('Label', group_keys=False).apply(lambda x: x.sample(1000))

  File "C:\Users\Msi\anaconda3\lib\site-packages\pandas\core\generic.py", line 5350, in sample
    locs = rs.choice(axis_length, size=n, replace=replace, p=weights)

  File "mtrand.pyx", line 959, in numpy.random.mtrand.RandomState.choice

ValueError: Cannot take a larger sample than population when 'replace=False'

CodePudding user response：

The messages means in, at least, one group you have not enough sample (< 1000). 2 solutions:

Use replace=True to get 1000 samples but some duplicates:

# You don't need apply here
Strat_d3 = d3.groupby('Label', group_keys=False).sample(1000, replace=True)

Use this trick if you accept some groups have less than 1000 samples:

Strat_d3 = d3.groupby('Label', group_keys=False).apply(lambda x: x.sample(min(len(x), 1000)))

To debug your groups, use the following code to check labels where number of samples are below 1000:

d3.value_counts('Label').loc[lambda x: x < 1000]