sorted crossval dataframe

src/enfobench/__version__.py

@@ -1 +1 @@
-__version__ = "0.3.4"
+__version__ = "0.3.5"

src/enfobench/dataset/utils.py

@@ -32,7 +32,8 @@ def create_perfect_forecasts_from_covariates(
         forecast = past_covariates.loc[(past_covariates.index > start) & (past_covariates.index <= start + horizon)]
         forecast.rename_axis("timestamp", inplace=True)
         forecast.reset_index(inplace=True)
-        forecast["cutoff_date"] = start.isoformat()  # pd.concat fails if cutoff_date is a Timestamp
+        forecast["cutoff_date"] = pd.to_datetime(start, unit="ns")
+        forecast.set_index(["cutoff_date", "timestamp"], inplace=True)
 
         if len(forecast) == 0:
             warnings.warn(
@@ -45,5 +46,6 @@ def create_perfect_forecasts_from_covariates(
         start += step
 
     forecast_df = pd.concat(forecasts, ignore_index=False)
-    forecast_df["cutoff_date"] = pd.to_datetime(forecast_df["cutoff_date"])  # convert back to Timestamp
+    forecast_df.reset_index(inplace=True)
+    forecast_df.sort_values(by=["cutoff_date", "timestamp"], inplace=True)
     return forecast_df

src/enfobench/evaluation/evaluate.py

@@ -159,4 +159,5 @@ def cross_validate(
 
     # Merge the forecast with the target
     crossval_df = crossval_df.merge(dataset._target, left_on="timestamp", right_index=True)
+    crossval_df.sort_values(by=["cutoff_date", "timestamp"], inplace=True)
     return crossval_df

tests/test_dataset/test_utils.py

@@ -11,6 +11,7 @@ def test_create_perfect_forecasts_from_covariates():
         data=np.random.rand(len(index), 2),
         columns=["covariate_1", "covariate_2"],
     )
+    past_covariates.drop(index[2], inplace=True)
 
     future_covariates = create_perfect_forecasts_from_covariates(
         past_covariates,