[Minor] Move max_lags, prediction_freq to config_model and n_lags to …

…config_ar (#1644) * updated dataset get_item * fixed linting issues * make targets contiguous * fixed ruff warnings * Unpack incrementally when needed * adjust forecaster * separate unpacking logic * added featureExtractor class * separate packing logic * fixed liniting issues * fixed covariates * remove lagged_reg_layers from model_config * remove n_lags from forecaster * remove model.n_lags references * fix typo * fix 2 * fixes * do not init max_lags * set max lags in add_lagged_reg * fix test * fix testz * fix testz2 * fix predict_seasonality * improve predic_seasonal_components * uncomment None configs * save previous settings * fix failing test * move prediction_frequency to model_config * remove lagged layers merge issue * remove packer * rm Extractor * rm Extractor2 * fix stacker in forecaster * fix test * fix tests * remove commented code * cleanup * ruff * remove unused fourier funcs and document new func * fix new func * retain OG fourier * move fourier helper to plotting utils * ruff --------- Co-authored-by: MaiBe-ctrl <[email protected]> Co-authored-by: Maisa Ben Salah <[email protected]>
ourownstory · Sep 3, 2024 · ca0952f · ca0952f
1 parent 9c4963c
commit ca0952f
Show file tree

Hide file tree

Showing 13 changed files with 241 additions and 206 deletions.
diff --git a/neuralprophet/configure.py b/neuralprophet/configure.py
@@ -5,7 +5,7 @@
 import types
 from collections import OrderedDict
 from dataclasses import dataclass, field
-from typing import Callable, List, Optional
+from typing import Callable, Dict, List, Optional
 from typing import OrderedDict as OrderedDictType
 from typing import Type, Union
 
@@ -23,8 +23,10 @@
 @dataclass
 class Model:
     features_map: dict
-    lagged_reg_layers: Optional[List[int]]
     quantiles: Optional[List[float]] = None
+    prediction_frequency: Optional[Dict[str]] = None
+    features_map: Optional[dict] = field(default_factory=dict)
+    max_lags: Optional[int] = field(init=False)
 
     def setup_quantiles(self):
         # convert quantiles to empty list [] if None
@@ -43,6 +45,32 @@ def setup_quantiles(self):
         # 0 is the median quantile index
         self.quantiles.insert(0, 0.5)
 
+    def set_max_num_lags(self, n_lags: int, config_lagged_regressors: Optional[ConfigLaggedRegressors] = None) -> int:
+        """Get the greatest number of lags between the autoregression lags and the covariates lags.
+
+        Parameters
+        ----------
+            n_lags : int
+                number of autoregressive lagged values of series to include as model inputs
+            config_lagged_regressors : configure.ConfigLaggedRegressors
+                Configurations for lagged regressors
+
+        Returns
+        -------
+            int
+                Maximum number of lags between the autoregression lags and the covariates lags.
+        """
+        if (
+            config_lagged_regressors is not None
+            and config_lagged_regressors.regressors is not None
+            and len(config_lagged_regressors.regressors) > 0
+        ):
+            lagged_regressor_lags = [val.n_lags for key, val in config_lagged_regressors.regressors.items()]
+            max_lagged_regressor_lags = max(lagged_regressor_lags)
+            self.max_lags = max(n_lags, max_lagged_regressor_lags)
+        else:
+            self.max_lags = n_lags
+
 
 ConfigModel = Model
 

diff --git a/neuralprophet/data/process.py b/neuralprophet/data/process.py
@@ -276,12 +276,12 @@ def _prepare_dataframe_to_predict(model, df: pd.DataFrame, max_lags: int, freq:
                 raise ValueError("only datestamps provided but y values needed for auto-regression.")
             df_i = _check_dataframe(model, df_i, check_y=False, exogenous=False)
         else:
-            df_i = _check_dataframe(model, df_i, check_y=model.max_lags > 0, exogenous=False)
+            df_i = _check_dataframe(model, df_i, check_y=model.config_model.max_lags > 0, exogenous=False)
             # fill in missing nans except for nans at end
             df_i = _handle_missing_data(
                 df=df_i,
                 freq=freq,
-                n_lags=model.n_lags,
+                n_lags=model.config_ar.n_lags,
                 n_forecasts=model.n_forecasts,
                 config_missing=model.config_missing,
                 config_regressors=model.config_regressors,
@@ -401,7 +401,7 @@ def _check_dataframe(
         pd.DataFrame
             checked dataframe
     """
-    if len(df) < (model.n_forecasts + model.n_lags) and not future:
+    if len(df) < (model.n_forecasts + model.config_ar.n_lags) and not future:
         raise ValueError(
             "Dataframe has less than n_forecasts + n_lags rows. "
             "Forecasting not possible. Please either use a larger dataset, or adjust the model parameters."
@@ -616,7 +616,7 @@ def _create_dataset(model, df, predict_mode, prediction_frequency=None, componen
     return time_dataset.GlobalTimeDataset(
         df,
         predict_mode=predict_mode,
-        n_lags=model.n_lags,
+        n_lags=model.config_ar.n_lags,
         n_forecasts=model.n_forecasts,
         prediction_frequency=prediction_frequency,
         predict_steps=model.predict_steps,

diff --git a/neuralprophet/df_utils.py b/neuralprophet/df_utils.py
@@ -88,29 +88,6 @@ def return_df_in_original_format(df, received_ID_col=False, received_single_time
     return new_df
 
 
-def get_max_num_lags(n_lags: int, config_lagged_regressors: Optional[ConfigLaggedRegressors]) -> int:
-    """Get the greatest number of lags between the autoregression lags and the covariates lags.
-
-    Parameters
-    ----------
-        n_lags : int
-            number of lagged values of series to include as model inputs
-        config_lagged_regressors : configure.ConfigLaggedRegressors
-            Configurations for lagged regressors
-
-    Returns
-    -------
-        int
-            Maximum number of lags between the autoregression lags and the covariates lags.
-    """
-    if config_lagged_regressors is not None and config_lagged_regressors.regressors is not None:
-        # log.debug("config_lagged_regressors exists")
-        return max([n_lags] + [val.n_lags for key, val in config_lagged_regressors.regressors.items()])
-    else:
-        # log.debug("config_lagged_regressors.regressors does not exist")
-        return n_lags
-
-
 def merge_dataframes(df: pd.DataFrame) -> pd.DataFrame:
     """Join dataframes for procedures such as splitting data, set auto seasonalities, and others.