Repository navigation
fix(revenue_tree): a single group loses its group label #602
Description
What: RevenueTree with group_col drops the group label when the data has exactly one group. The result gets a RangeIndex instead of store=Downtown. Checked on main @ c5f4b2a.
Cause: revenue_tree.py:70-73 in calc_tree_kpis resets the index when len(df.index) == period_count. That condition is meant to detect "no group_col", but one group also aggregates to two rows. The same reset fires with two groups when one is missing from a period. The decision should rest on whether group_col is set, not on the row count.
Failing test:
import pandas as pd
from openretailscience.analysis.revenue_tree import RevenueTree
from openretailscience.options import ColumnHelper
cols = ColumnHelper()
class TestRevenueTreeGroupLabels:
def test_single_group_keeps_its_label(self):
"""A single-store analysis (two aggregated rows) is still indexed by store."""
downtown = pd.DataFrame(
{
cols.customer_id: [101, 102, 101],
cols.transaction_id: [9001, 9002, 9004],
cols.unit_spend: [20.50, 30.00, 40.00],
"quarter": ["2024Q1", "2024Q1", "2024Q2"],
"store": ["Downtown", "Downtown", "Downtown"],
},
)
result = RevenueTree(
df=downtown, period_col="quarter", p1_value="2024Q1", p2_value="2024Q2", group_col="store"
).df
assert result.index.name == "store"
assert list(result.index) == ["Downtown"]
Activity