Skip to content

BUG: Require sample weights to sum to less than 1 when replace = True #61582

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 3 commits into
base: main
Choose a base branch
from
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions doc/source/whatsnew/v3.0.0.rst
Original file line number Diff line number Diff line change
Expand Up @@ -888,6 +888,7 @@ Other
- Bug in :meth:`DataFrame.query` where using duplicate column names led to a ``TypeError``. (:issue:`59950`)
- Bug in :meth:`DataFrame.query` which raised an exception or produced incorrect results when expressions contained backtick-quoted column names containing the hash character ``#``, backticks, or characters that fall outside the ASCII range (U+0001..U+007F). (:issue:`59285`) (:issue:`49633`)
- Bug in :meth:`DataFrame.query` which raised an exception when querying integer column names using backticks. (:issue:`60494`)
- Bug in :meth:`DataFrame.sample` with ``replace=False`` and ``(n * max(weights) / sum(weights)) > 1``, the method would return biased results. Now raises ``ValueError``. (:issue:`61516`)
- Bug in :meth:`DataFrame.shift` where passing a ``freq`` on a DataFrame with no columns did not shift the index correctly. (:issue:`60102`)
- Bug in :meth:`DataFrame.sort_index` when passing ``axis="columns"`` and ``ignore_index=True`` and ``ascending=False`` not returning a :class:`RangeIndex` columns (:issue:`57293`)
- Bug in :meth:`DataFrame.sort_values` where sorting by a column explicitly named ``None`` raised a ``KeyError`` instead of sorting by the column as expected. (:issue:`61512`)
Expand Down
2 changes: 2 additions & 0 deletions pandas/core/generic.py
Original file line number Diff line number Diff line change
Expand Up @@ -5815,6 +5815,8 @@ def sample(
If weights do not sum to 1, they will be normalized to sum to 1.
Missing values in the weights column will be treated as zero.
Infinite values not allowed.
When replace = False will not allow ``(n * max(weights) / sum(weights)) > 1``,
in order to avoid biased results.
random_state : int, array-like, BitGenerator, np.random.RandomState, np.random.Generator, optional
If int, array-like, or BitGenerator, seed for random number generator.
If np.random.RandomState or np.random.Generator, use as given.
Expand Down
8 changes: 8 additions & 0 deletions pandas/core/sample.py
Original file line number Diff line number Diff line change
Expand Up @@ -150,6 +150,14 @@ def sample(
else:
raise ValueError("Invalid weights: weights sum to zero")


is_max_weight_dominating = size * max(weights) > weight_sum
if is_max_weight_dominating and not replace:
raise ValueError(
"Invalid weights: If `replace`=False,"
" unit probabilities have to be less than 1"
)

return random_state.choice(obj_len, size=size, replace=replace, p=weights).astype(
np.intp, copy=False
)
11 changes: 11 additions & 0 deletions pandas/tests/frame/methods/test_sample.py
Original file line number Diff line number Diff line change
Expand Up @@ -137,6 +137,17 @@ def test_sample_inf_weights(self, obj):
with pytest.raises(ValueError, match=msg):
obj.sample(n=3, weights=weights_with_ninf)

def test_sample_replacement_weight_sum(self, obj):
# GH#61516
weights_large_total = [1] * 10
weights_large_total[0] = 100
msg = (
"Invalid weights: If `replace`=False,"
" unit probabilities have to be less than 1"
)
with pytest.raises(ValueError, match=msg):
obj.sample(n=3, weights=weights_large_total, replace=False)

def test_sample_zero_weights(self, obj):
# All zeros raises errors

Expand Down
Loading