Merge pull request #141 from wilsonrljr/check_polynomial

change Polynomial check from class name to isinstance method.

Author: wilsonrljr

pyproject.toml

@@ -57,6 +57,7 @@ dependencies = ["numpy>=1.19.2,<2.0", "scipy>=1.7.0", "matplotlib>=3.3.2"]
 dynamic = ["version"]
 
 [project.optional-dependencies]
+
 dev = [
     "pytest >=7.0.0,<8.0.0",
     "pytest-cov >=2.12.0,<4.0.0",
@@ -87,7 +88,7 @@ doc = [
     # avoid jinja import error using 3.0.3 version
     "jinja2==3.0.3",
 ]
-all = ["torch >=1.7.1"]
+all = ["torch >=1.7.1, <2.4.0"]
 
 [project.urls]
 homepage = "http://sysidentpy.org"

sysidentpy/general_estimators/narx.py

@@ -160,7 +160,12 @@ def fit(self, *, X=None, y=None):
         self.max_lag = self._get_max_lag()
         lagged_data = self.build_matrix(X, y)
         reg_matrix = self.basis_function.fit(
-            lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type, predefined_regressors=None
+            lagged_data,
+            self.max_lag,
+            self.ylag,
+            self.xlag,
+            self.model_type,
+            predefined_regressors=None,
         )
 
         if X is not None:
@@ -213,7 +218,7 @@ def predict(
             The predicted values of the model.
 
         """
-        if self.basis_function.__class__.__name__ == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             if steps_ahead is None:
                 yhat = self._model_prediction(X, y, forecast_horizon=forecast_horizon)
                 yhat = np.concatenate([y[: self.max_lag], yhat], axis=0)
@@ -263,11 +268,7 @@ def _one_step_ahead_prediction(self, X, y):
         """
         lagged_data = self.build_matrix(X, y)
         X_base = self.basis_function.transform(
-            lagged_data,
-            self.max_lag,
-            self.ylag,
-            self.xlag,
-            self.model_type
+            lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type
         )
 
         yhat = self.base_estimator.predict(X_base)
@@ -485,11 +486,7 @@ def _basis_function_predict(self, X, y_initial, forecast_horizon=None):
                 yhat[i : i + analyzed_elements_number].reshape(-1, 1),
             )
             X_tmp = self.basis_function.transform(
-                lagged_data,
-                self.max_lag,
-                self.ylag,
-                self.xlag,
-                self.model_type
+                lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type
             )
 
             a = self.base_estimator.predict(X_tmp)

sysidentpy/model_structure_selection/accelerated_orthogonal_least_squares.py

@@ -308,7 +308,12 @@ def fit(self, *, X: Optional[np.ndarray] = None, y: Optional[np.ndarray] = None)
         self.max_lag = self._get_max_lag()
         lagged_data = self.build_matrix(X, y)
         reg_matrix = self.basis_function.fit(
-            lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type, predefined_regressors=None
+            lagged_data,
+            self.max_lag,
+            self.ylag,
+            self.xlag,
+            self.model_type,
+            predefined_regressors=None,
         )
 
         if X is not None:
@@ -371,7 +376,7 @@ def predict(
             The predicted values of the model.
 
         """
-        if self.basis_function.__class__.__name__ == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             if steps_ahead is None:
                 yhat = self._model_prediction(X, y, forecast_horizon=forecast_horizon)
                 yhat = np.concatenate([y[: self.max_lag], yhat], axis=0)

sysidentpy/model_structure_selection/entropic_regression.py

@@ -584,7 +584,12 @@ def fit(self, *, X=None, y=None):
         lagged_data = self.build_matrix(X, y)
 
         reg_matrix = self.basis_function.fit(
-            lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type, predefined_regressors=None
+            lagged_data,
+            self.max_lag,
+            self.ylag,
+            self.xlag,
+            self.model_type,
+            predefined_regressors=None,
         )
 
         if X is not None:
@@ -687,7 +692,7 @@ def predict(self, *, X=None, y=None, steps_ahead=None, forecast_horizon=None):
             The predicted values of the model.
 
         """
-        if self.basis_function.__class__.__name__ == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             if steps_ahead is None:
                 yhat = self._model_prediction(X, y, forecast_horizon=forecast_horizon)
                 yhat = np.concatenate([y[: self.max_lag], yhat], axis=0)

sysidentpy/model_structure_selection/forward_regression_orthogonal_least_squares.py

@@ -602,7 +602,12 @@ def fit(self, *, X: Optional[np.ndarray] = None, y: np.ndarray):
         lagged_data = self.build_matrix(X, y)
 
         reg_matrix = self.basis_function.fit(
-            lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type, predefined_regressors=None
+            lagged_data,
+            self.max_lag,
+            self.ylag,
+            self.xlag,
+            self.model_type,
+            predefined_regressors=None,
         )
 
         if X is not None:
@@ -689,7 +694,7 @@ def predict(
             The predicted values of the model.
 
         """
-        if self.basis_function.__class__.__name__ == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             if steps_ahead is None:
                 yhat = self._model_prediction(X, y, forecast_horizon=forecast_horizon)
                 yhat = np.concatenate([y[: self.max_lag], yhat], axis=0)

sysidentpy/model_structure_selection/meta_model_structure_selection.py

@@ -385,7 +385,12 @@ def evaluate_objective_function(
             lagged_data = self.build_matrix(X_train, y_train)
 
             psi = self.basis_function.fit(
-                lagged_data, self.max_lag, self.xlag, self.ylag, self.model_type, predefined_regressors=self.pivv
+                lagged_data,
+                self.max_lag,
+                self.xlag,
+                self.ylag,
+                self.model_type,
+                predefined_regressors=self.pivv,
             )
 
             pos_insignificant_terms, _, _ = self.perform_t_test(
@@ -607,7 +612,7 @@ def predict(
             The predicted values of the model.
 
         """
-        if self.basis_function.__class__.__name__ == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             if steps_ahead is None:
                 yhat = self._model_prediction(X, y, forecast_horizon=forecast_horizon)
                 yhat = np.concatenate([y[: self.max_lag], yhat], axis=0)

sysidentpy/neural_network/narx_nn.py

@@ -271,15 +271,24 @@ def split_data(self, X, y):
         self.max_lag = self._get_max_lag()
         lagged_data = self.build_matrix(X, y)
 
-        basis_name = self.basis_function.__class__.__name__
-        if basis_name == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             reg_matrix = self.basis_function.fit(
-                lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type, predefined_regressors=None
+                lagged_data,
+                self.max_lag,
+                self.ylag,
+                self.xlag,
+                self.model_type,
+                predefined_regressors=None,
             )
             reg_matrix = reg_matrix[:, 1:]
         else:
             reg_matrix = self.basis_function.fit(
-                lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type, predefined_regressors=None
+                lagged_data,
+                self.max_lag,
+                self.ylag,
+                self.xlag,
+                self.model_type,
+                predefined_regressors=None,
             )
 
         if X is not None:
@@ -476,7 +485,7 @@ def predict(self, *, X=None, y=None, steps_ahead=None, forecast_horizon=None):
             The predicted values of the model.
 
         """
-        if self.basis_function.__class__.__name__ == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             if steps_ahead is None:
                 return self._model_prediction(X, y, forecast_horizon=forecast_horizon)
             if steps_ahead == 1:
@@ -513,23 +522,14 @@ def _one_step_ahead_prediction(self, X, y):
         """
         lagged_data = self.build_matrix(X, y)
 
-        basis_name = self.basis_function.__class__.__name__
-        if basis_name == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             X_base = self.basis_function.transform(
-                lagged_data,
-                self.max_lag,
-                self.ylag,
-                self.xlag,
-                self.model_type
+                lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type
             )
             X_base = X_base[:, 1:]
         else:
             X_base = self.basis_function.transform(
-                lagged_data,
-                self.max_lag,
-                self.ylag,
-                self.xlag,
-                self.model_type
+                lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type
             )
 
         yhat = np.zeros(X.shape[0], dtype=float)
@@ -716,11 +716,7 @@ def _basis_function_predict(self, X, y_initial, forecast_horizon=None):
                 )
 
             X_tmp = self.basis_function.transform(
-                lagged_data,
-                self.max_lag,
-                self.ylag,
-                self.xlag,
-                self.model_type
+                lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type
             )
             X_tmp = np.atleast_1d(X_tmp).astype(np.float32)
             yhat = yhat.astype(np.float32)

sysidentpy/simulation/_simulation.py

@@ -179,7 +179,7 @@ def _validate_simulate_params(self):
             )
 
     def _check_simulate_params(self, y_train, y_test, model_code, steps_ahead, theta):
-        if self.basis_function.__class__.__name__ != "Polynomial":
+        if not isinstance(self.basis_function, Polynomial):
             raise NotImplementedError(
                 "Currently, SimulateNARMAX only works for polynomial models."
             )
@@ -277,7 +277,12 @@ def simulate(
             self.max_lag = self._get_max_lag()
             lagged_data = self.build_matrix(X_train, y_train)
             psi = self.basis_function.fit(
-                lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type, predefined_regressors=self.pivv
+                lagged_data,
+                self.max_lag,
+                self.ylag,
+                self.xlag,
+                self.model_type,
+                predefined_regressors=self.pivv,
             )
 
             self.theta = self.estimator.optimize(
@@ -303,7 +308,12 @@ def simulate(
             self.max_lag = self._get_max_lag()
             lagged_data = self.build_matrix(X_train, y_train)
             psi = self.basis_function.fit(
-                lagged_data, self.max_lag, self.ylag, self.xlag, self.model_type, predefined_regressors=self.pivv
+                lagged_data,
+                self.max_lag,
+                self.ylag,
+                self.xlag,
+                self.model_type,
+                predefined_regressors=self.pivv,
             )
 
             _, self.err, _, _ = self.error_reduction_ratio(
@@ -434,7 +444,7 @@ def predict(self, *, X=None, y=None, steps_ahead=None, forecast_horizon=None):
             The predicted values of the model.
 
         """
-        if self.basis_function.__class__.__name__ == "Polynomial":
+        if isinstance(self.basis_function, Polynomial):
             if steps_ahead is None:
                 yhat = self._model_prediction(X, y, forecast_horizon=forecast_horizon)
                 yhat = np.concatenate([y[: self.max_lag], yhat], axis=0)

sysidentpy/utils/narmax_tools.py

@@ -4,6 +4,7 @@
 import numpy as np
 
 from ..narmax_base import RegressorDictionary
+from ..basis_function import Polynomial
 from ._check_arrays import _num_features
 
 
@@ -47,24 +48,28 @@ def regressor_code(
         xlag=xlag, ylag=ylag, model_type=model_type, basis_function=basis_function
     ).regressor_space(n_inputs)
 
-    basis_name = basis_function.__class__.__name__
-    if basis_name != "Polynomial" and basis_function.ensemble:
+    if not isinstance(basis_function, Polynomial) and basis_function.ensemble:
         repetition = basis_function.n * 2
         basis_code = np.sort(
             np.tile(encoding[1:, :], (repetition, 1)),
             axis=0,
         )
         encoding = np.concatenate([encoding[1:], basis_code])
-    elif basis_name != "Polynomial" and basis_function.ensemble is False:
+    elif (
+        not isinstance(basis_function, Polynomial) and basis_function.ensemble is False
+    ):
         repetition = basis_function.n * 2
         encoding = np.sort(
             np.tile(encoding[1:, :], (repetition, 1)),
             axis=0,
         )
 
-    if basis_name == "Polynomial" and model_representation == "neural_network":
+    if (
+        isinstance(basis_function, Polynomial)
+        and model_representation == "neural_network"
+    ):
         return encoding[1:]
-    if basis_name == "Polynomial" and model_representation is None:
+    if isinstance(basis_function, Polynomial) and model_representation is None:
         return encoding
 
     return encoding