Skip to content

Fix partition_dataset losing items when rounded ratio sizes fall short #9230

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.

Sign up for GitHub

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

Open
rtmalikian wants to merge 1 commit into
base: dev
Choose a base branch
from
Open

Fix partition_dataset losing items when rounded ratio sizes fall short #9230

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
11 changes: 9 additions & 2 deletions monai/data/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1203,9 +1203,16 @@ def partition_dataset(
if ratios:
next_idx = 0
rsum = sum(ratios)
for r in ratios:
sizes = [int(r / rsum * data_len + 0.5) for r in ratios]
if sum(sizes) < data_len:
# Independent rounding can leave trailing items in no partition
# (e.g. ratios [1, 1, 1] on 10 items rounds to 3 + 3 + 3); assign
# the remainder to the last partition so every item lands in
# exactly one partition.
sizes[-1] += data_len - sum(sizes)
for size in sizes:
start_idx = next_idx
next_idx = min(start_idx + int(r / rsum * data_len + 0.5), data_len)
next_idx = min(start_idx + size, data_len)
datasets.append([data[i] for i in indices[start_idx:next_idx]])
return datasets

Expand Down
69 changes: 67 additions & 2 deletions tests/data/test_partition_dataset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -117,10 +117,75 @@
]


class TestPartitionDataset(unittest.TestCase):
TEST_CASE_9 = [
{
"data": list(range(10)),
"ratios": [1, 1, 1],
"num_partitions": None,
"shuffle": False,
"seed": 0,
"drop_last": True,
"even_divisible": True,
},
[[0, 1, 2], [3, 4, 5], [6, 7, 8, 9]],
]

TEST_CASE_10 = [
{
"data": list(range(63)),
"ratios": [0.8, 0.1, 0.1],
"num_partitions": None,
"shuffle": False,
"seed": 0,
"drop_last": True,
"even_divisible": True,
},
[list(range(50)), list(range(50, 56)), list(range(56, 63))],
]

TEST_CASE_11 = [
{
"data": list(range(32)),
"ratios": [1, 1, 1, 1, 1],
"num_partitions": None,
"shuffle": False,
"seed": 0,
"drop_last": True,
"even_divisible": True,
},
[list(range(6)), list(range(6, 12)), list(range(12, 18)), list(range(18, 24)), list(range(24, 32))],
]

TEST_CASE_12 = [
{
"data": [1, 2, 3, 4, 5],
"ratios": [0.6, 0.2, 0.2],
"num_partitions": None,
"shuffle": False,
"seed": 0,
"drop_last": True,
"even_divisible": True,
},
[[1, 2, 3], [4], [5]],
]


class TestPartitionDataset(unittest.TestCase):
@parameterized.expand(
[TEST_CASE_1, TEST_CASE_2, TEST_CASE_3, TEST_CASE_4, TEST_CASE_5, TEST_CASE_6, TEST_CASE_7, TEST_CASE_8]
[
TEST_CASE_1,
TEST_CASE_2,
TEST_CASE_3,
TEST_CASE_4,
TEST_CASE_5,
TEST_CASE_6,
TEST_CASE_7,
TEST_CASE_8,
TEST_CASE_9,
TEST_CASE_10,
TEST_CASE_11,
TEST_CASE_12,
]
)
def test_value(self, input_param, result):
self.assertListEqual(partition_dataset(**input_param), result)
Expand Down
Loading