Skip to content

Ignore unused categorical IDs during feature extraction #1137

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
AryanRG wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Ignore unused categorical IDs during feature extraction #1137

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
43 changes: 43 additions & 0 deletions tests/units/feature_extraction/test_categorical_ids.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,43 @@
"""Feature extraction must not invent time series for unused ID categories."""

import pandas as pd
import pytest

from tsfresh import extract_features


@pytest.mark.parametrize("data_format", ["wide", "long", "dict"])
@pytest.mark.parametrize("column_sort", [None, "time"])
@pytest.mark.parametrize("id_type", ["integer", "categorical", "unused_category"])
def test_extract_features_only_observed_ids(data_format, column_sort, id_type):
"""Filtering categorical IDs must produce the same rows as integer IDs."""
df = pd.DataFrame({"id": [1, 3, 1, 3], "time": [2, 2, 1, 1], "value": [4, 8, 2, 6]})
if id_type != "integer":
categories = [1, 2, 3] if id_type == "unused_category" else [1, 3]
df["id"] = pd.Categorical(df["id"], categories=categories)

kwargs = {}
if data_format == "dict":
data = {"value": df}
kwargs["column_value"] = "value"
elif data_format == "long":
data = df.assign(kind="value")
kwargs.update(column_kind="kind", column_value="value")
else:
data = df
kwargs["column_value"] = "value"

result = extract_features(
data,
column_id="id",
column_sort=column_sort,
default_fc_parameters={"length": None, "sum_values": None},
n_jobs=1,
disable_progressbar=True,
**kwargs,
)
expected = pd.DataFrame(
{"value__length": [2.0, 2.0], "value__sum_values": [6.0, 14.0]},
index=pd.Index([1, 3], dtype=df["id"].dtype),
)
pd.testing.assert_frame_equal(result, expected)
6 changes: 3 additions & 3 deletions tsfresh/feature_extraction/data.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -214,7 +214,7 @@ def __init__(self, df, column_id, column_sort=None, value_columns=None):
_check_nan(df, column_sort)

self.column_sort = column_sort
self.df_grouped = df.groupby(column_id)
self.df_grouped = df.groupby(column_id, observed=True)

super().__init__(df, column_id)

Expand Down Expand Up @@ -319,12 +319,12 @@ def __init__(self, ts_dict, column_id, column_value, column_sort=None):
_check_nan(df, column_sort)

self.grouped_dict = {
key: df.sort_values([column_sort]).groupby(column_id)
key: df.sort_values([column_sort]).groupby(column_id, observed=True)
for key, df in ts_dict.items()
}
else:
self.grouped_dict = {
key: df.groupby(column_id) for key, df in ts_dict.items()
key: df.groupby(column_id, observed=True) for key, df in ts_dict.items()
}

super().__init__(df, column_id)
Expand Down