Revert "Allow manually setting 'freq' for specific datasets with simp…

…le data …"

This reverts commit 901a987.

src/uni2ts/data/builder/simple.py

@@ -37,22 +37,9 @@ def _from_long_dataframe(
     df: pd.DataFrame,
     offset: Optional[int] = None,
     date_offset: Optional[pd.Timestamp] = None,
-    freq: str = "H",
 ) -> tuple[GenFunc, Features]:
     items = df.item_id.unique()
 
-    # Infer the freq and generate the prompt
-    inferred_freq = pd.infer_freq(df.index)
-
-    if inferred_freq is not None:
-        print(
-            f"Inferred frequency: {inferred_freq}. Using this value for the 'freq' parameter."
-        )
-    else:
-        print(
-            f"Inferred frequency is None. Using predefined {freq} for the 'freq' parameter."
-        )
-
     def example_gen_func() -> Generator[dict[str, Any], None, None]:
         for item_id in items:
             item_df = df.query(f'item_id == "{item_id}"').drop("item_id", axis=1)
@@ -63,11 +50,7 @@ def example_gen_func() -> Generator[dict[str, Any], None, None]:
             yield {
                 "target": item_df.to_numpy(),
                 "start": item_df.index[0],
-                "freq": (
-                    pd.infer_freq(df.index)
-                    if pd.infer_freq(df.index) is not None
-                    else freq
-                ),
+                "freq": pd.infer_freq(item_df.index),
                 "item_id": item_id,
             }
 
@@ -87,7 +70,6 @@ def _from_wide_dataframe(
     df: pd.DataFrame,
     offset: Optional[int] = None,
     date_offset: Optional[pd.Timestamp] = None,
-    freq: str = "H",
 ) -> tuple[GenFunc, Features]:
     if offset is not None:
         df = df.iloc[:offset]
@@ -96,28 +78,12 @@ def _from_wide_dataframe(
 
     print(df)
 
-    # Infer the freq and generate the prompt
-    inferred_freq = pd.infer_freq(df.index)
-
-    if inferred_freq is not None:
-        print(
-            f"Inferred frequency: {inferred_freq}. Using this value for the 'freq' parameter."
-        )
-    else:
-        print(
-            f"Inferred frequency is None. Using predefined {freq} for the 'freq' parameter."
-        )
-
     def example_gen_func() -> Generator[dict[str, Any], None, None]:
         for i in range(len(df.columns)):
             yield {
                 "target": df.iloc[:, i].to_numpy(),
                 "start": df.index[0],
-                "freq": (
-                    pd.infer_freq(df.index)
-                    if pd.infer_freq(df.index) is not None
-                    else freq
-                ),
+                "freq": pd.infer_freq(df.index),
                 "item_id": f"item_{i}",
             }
 
@@ -137,32 +103,17 @@ def _from_wide_dataframe_multivariate(
     df: pd.DataFrame,
     offset: Optional[int] = None,
     date_offset: Optional[pd.Timestamp] = None,
-    freq: str = "H",
 ) -> tuple[GenFunc, Features]:
     if offset is not None:
         df = df.iloc[:offset]
     elif date_offset is not None:
         df = df[df.index <= date_offset]
 
-    # Infer the freq and generate the prompt
-    inferred_freq = pd.infer_freq(df.index)
-
-    if inferred_freq is not None:
-        print(
-            f"Inferred frequency: {inferred_freq}. Using this value for the 'freq' parameter."
-        )
-    else:
-        print(
-            f"Inferred frequency is None. Using predefined {freq} for the 'freq' parameter."
-        )
-
     def example_gen_func() -> Generator[dict[str, Any], None, None]:
         yield {
             "target": df.to_numpy().T,
             "start": df.index[0],
-            "freq": (
-                pd.infer_freq(df.index) if pd.infer_freq(df.index) is not None else freq
-            ),
+            "freq": pd.infer_freq(df.index),
             "item_id": "item_0",
         }
 
@@ -194,7 +145,6 @@ def build_dataset(
         dataset_type: str,
         offset: Optional[int] = None,
         date_offset: Optional[pd.Timestamp] = None,
-        freq: str = "H",
     ):
         assert offset is None or date_offset is None, (
             "One or neither offset and date_offset must be specified, but not both. "
@@ -216,7 +166,7 @@ def build_dataset(
             )
 
         example_gen_func, features = _from_dataframe(
-            df, freq=freq, offset=offset, date_offset=date_offset
+            df, offset=offset, date_offset=date_offset
         )
         hf_dataset = datasets.Dataset.from_generator(
             example_gen_func, features=features
@@ -253,7 +203,7 @@ class SimpleEvalDatasetBuilder(DatasetBuilder):
     def __post_init__(self):
         self.storage_path = Path(self.storage_path)
 
-    def build_dataset(self, file: Path, dataset_type: str, freq: str = "H"):
+    def build_dataset(self, file: Path, dataset_type: str):
         df = pd.read_csv(file, index_col=0, parse_dates=True)
 
         if dataset_type == "long":
@@ -268,7 +218,7 @@ def build_dataset(self, file: Path, dataset_type: str, freq: str = "H"):
                 " Valid options are 'long', 'wide', and 'wide_multivariate'."
             )
 
-        example_gen_func, features = _from_dataframe(df, freq=freq)
+        example_gen_func, features = _from_dataframe(df)
         hf_dataset = datasets.Dataset.from_generator(
             example_gen_func, features=features
         )
@@ -339,21 +289,13 @@ def generate_eval_builders(
         type=str,
         default=None,
     )
-    # Define the `freq` argument with a default value. Use this value as 'freq' if 'freq' is None.
-    parser.add_argument(
-        "--freq",
-        default="H",  # Set the default value
-        help="The user specified frequency",
-    )
-
     args = parser.parse_args()
 
     SimpleDatasetBuilder(dataset=args.dataset_name).build_dataset(
         file=Path(args.file_path),
         dataset_type=args.dataset_type,
         offset=args.offset,
         date_offset=pd.Timestamp(args.date_offset) if args.date_offset else None,
-        freq=args.freq,
     )
 
     if args.offset is not None or args.date_offset is not None:
@@ -365,6 +307,4 @@ def generate_eval_builders(
             prediction_length=None,
             context_length=None,
             patch_size=None,
-        ).build_dataset(
-            file=Path(args.file_path), dataset_type=args.dataset_type, freq=args.freq
-        )
+        ).build_dataset(file=Path(args.file_path), dataset_type=args.dataset_type)