Merge pull request #517 from openeemeter/fix/billing-dst

Handle ambiguous and nonexistent local times when creating billing data

CHANGELOG.md

@@ -4,6 +4,7 @@ Changelog
 Development
 -----------
 
+* Handle ambiguous and nonexistent local times when creating billing dataclass 
 * Fix serialization and deserialization of metrics.
 * Rename HourlyBaselineData.sufficiency_warnings -> HourlyBaselineData.warnings
 * Add disqualification field to HourlyBaselineData and HourlyReportingData

eemeter/eemeter/models/billing/data.py

@@ -125,6 +125,8 @@ def _compute_meter_value_df(self, df: pd.DataFrame):
                 end=end_date,
                 freq="D",
                 tz=df.index.tz,
+                ambiguous=True,
+                nonexistent="shift_forward",
             )
             all_days_df = pd.DataFrame(index=all_days_index)
             meter_value_df = meter_value_df.merge(

tests/daily_model/test_billing_data.py

@@ -24,6 +24,7 @@
 from eemeter.eemeter.samples import load_sample
 import numpy as np
 import pandas as pd
+from pandas import Timestamp, DatetimeIndex, DataFrame
 import pytest
 
 TEMPERATURE_SEED = 29
@@ -634,3 +635,21 @@ def test_billing_reporting_data_with_missing_daily_frequencies(get_datetime_inde
         disqualification.qualified_name in expected_disqualifications
         for disqualification in cls.disqualification
     )
+
+
+def test_dst_handling():
+    # 2020-03-08 02:00 is nonexistent, should push to 03:00
+    tz = "America/New_York"
+    idx = DatetimeIndex([Timestamp("2020-03-07 02", tz=tz), Timestamp("2020-04-06 02", tz=tz), Timestamp("2020-05-06 02", tz=tz)])
+    df = DataFrame({"observed": [1]*3, "temperature": [50]*3}, index=idx)
+    baseline = BillingBaselineData(df, is_electricity_data=True)
+    assert len(baseline.df) == 61
+    hours = np.unique(baseline.df.index.hour)
+    assert (hours == [2, 3]).all()
+
+    # 2020-11-01 01:00 is ambiguous, single index should be chosen
+    tz = "America/New_York"
+    idx = DatetimeIndex([Timestamp("2020-10-31 01", tz=tz), Timestamp("2020-11-28 01", tz=tz), Timestamp("2020-12-28 01", tz=tz)])
+    df = DataFrame({"observed": [1]*3, "temperature": [50]*3}, index=idx)
+    baseline = BillingBaselineData(df, is_electricity_data=True)
+    assert (baseline.df.index.hour == 1).all()