predict for model implemented

03_DataPostProcessing.py

@@ -5,7 +5,7 @@ from xml.etree import ElementTree as ET
 from sklearn.linear_model import LinearRegression
 from sklearn.preprocessing import StandardScaler, MinMaxScaler
 import joblib
-from miscellaneous.DataPreparation import load_xml_to_numpy, remove_rows
+from miscellaneous.DataPreparation import load_xml_to_numpy, remove_rows_abstract, remove_rows, remove_rows_iloc
 import os
 import shutil
 
@@ -15,7 +15,7 @@ VELOCITY_SIZE = 3840
 PRESSURE_SIZE = 1080
 
 for stage in ['train', 'test']:
-    df = pd.read_excel(f'Data/{stage}/Parameter_input.xlsx').transpose()
+    df = pd.read_excel(f'Data/{stage}/parameter_input.xlsx').transpose()
     N_FILES_STAGE = df.shape[1]
 
     # Initialize empty arrays
@@ -26,13 +26,17 @@ for stage in ['train', 'test']:
         file_identifier = f'{index}.xml'
 
         velocity_file = f'Data/{stage}/velocities/velocity_field_{file_identifier}'
-        pressure_file = f'Data/{stage}/Pressure/pressure_field_{file_identifier}'
+        pressure_file = f'Data/{stage}/pressure/pressure_field_{file_identifier}'
 
         velocity_matrix[:,index] = load_xml_to_numpy(velocity_file)
         pressure_matrix[:,index] = load_xml_to_numpy(pressure_file)
 
-    velocity_matrix, velocity_removed_rows_identifier = remove_rows(velocity_matrix, atol=1e-4, remove_nan=True)
-    pressure_matrix, pressure_removed_rows_identifier = remove_rows(pressure_matrix, atol=1e-4, remove_nan=True)
+    if stage == 'train':
+        velocity_matrix, velocity_removed_rows_identifier = remove_rows(velocity_matrix, 1e-4, True)
+        pressure_matrix, pressure_removed_rows_identifier = remove_rows(pressure_matrix, 1e-4, True)
+    elif stage == 'test':
+        velocity_matrix = remove_rows_iloc(velocity_matrix, velocity_removed_rows_identifier)
+        pressure_matrix = remove_rows_iloc(pressure_matrix, velocity_removed_rows_identifier)
 
     # Save the data
     if os.path.exists(f'Data/{stage}/matrices'):

04_TrainModels.py

@@ -6,4 +6,8 @@ regMod = RegressionModels()
 linregMod = LinearRegressionModel()
 
 X_r_velocity, X_r_pressure = linregMod.POD(10)
-model_velocity, model_pressure = linregMod.train(X_r_velocity, X_r_pressure)
\ No newline at end of file
+model_velocity, model_pressure = linregMod.train(X_r_velocity, X_r_pressure)
+
+predict_velocity_test, predict_pressure_test = linregMod.predict_test(model_velocity, model_pressure, scale=False)
+
+predict_velocity_test_rescaled, predict_pressure_test_rescaled = linregMod.predict_test(model_velocity, model_pressure, scale=True)

miscellaneous/DataPreparation.py

@@ -15,7 +15,11 @@ def load_xml_to_numpy(file_path):
 
     return data
 
-
+def remove_rows_abstract(matrix, atol, remove_nan, stage, removed_rows_identifier):
+    if stage == 'train':
+        return remove_rows(matrix=matrix, atol=atol, remove_nan=remove_nan)
+    elif stage == 'test':
+        return remove_rows_iloc(matrix=matrix, removed_rows_identifier=removed_rows_identifier, remove_nan=remove_nan)
 
 def remove_rows(matrix, atol=1e-8, remove_nan=True):
     i = 0
@@ -30,6 +34,20 @@ def remove_rows(matrix, atol=1e-8, remove_nan=True):
         i += 1
         removed_rows_counter += 1
 
-    matrix = np.nan_to_num(matrix, copy=False, nan=0.0, posinf=0.0, neginf=0.0)
+    if remove_nan:
+        matrix = np.nan_to_num(matrix, copy=False, nan=0.0, posinf=0.0, neginf=0.0)
+
+    return matrix, np.array(removed_rows_identifier)
+
+def remove_rows_iloc(matrix, removed_rows_identifier, remove_nan=True):
+    i = 0
+    while i < matrix.shape[0]:
+        if i in removed_rows_identifier:
+            matrix = np.concatenate((matrix[:i,:], matrix[i+1:,:]), axis=0)
+            continue
+        i += 1
+
+    if remove_nan:
+        matrix = np.nan_to_num(matrix, copy=False, nan=0.0, posinf=0.0, neginf=0.0)
 
-    return matrix, np.array(removed_rows_identifier)
\ No newline at end of file
+    return matrix
\ No newline at end of file

models/LinearRegression.py

@@ -16,7 +16,5 @@ class LinearRegressionModel(RegressionModels):
 
         return model_velocity, model_pressure
     
-    # def predict(self, input, model_velocity, model_pressure, scaled=True):
-    #     predict_velocity = model_velocity.predict(, self.param_matrix_test, scaled)
-
-    #     return model.predict(X_test)
\ No newline at end of file
+    def predict_test(self, model_velocity, model_pressure, scale=True):
+        return super().predict_test(model_velocity, model_pressure, scale)
\ No newline at end of file

models/RegressionModels.py

@@ -49,4 +49,21 @@ class RegressionModels:
         X_r_velocity = U_r_velocity @ (S_r_velocity @ V_r_velocity)
         X_r_pressure = U_r_pressure @ (S_r_pressure @ V_r_pressure)
 
-        return X_r_velocity, X_r_pressure
\ No newline at end of file
+        return X_r_velocity, X_r_pressure
+    
+    def scale_(self, data_velocity, data_pressure):
+        return self.scaler_velocity.transform(data_velocity), self.scaler_pressure.transform(data_pressure)
+
+    def inverse_scale_(self, data_velocity, data_pressure):
+        return self.scaler_velocity.inverse_transform(data_velocity), self.scaler_pressure.inverse_transform(data_pressure)
+    
+    def predict_test(self, model_velocity, model_pressure, scale=True):
+        prediction_velocity = model_velocity.predict(self.param_matrix_test)
+        prediction_pressure = model_pressure.predict(self.param_matrix_test)
+        if scale==True:
+            prediction_velocity, prediction_pressure = self.inverse_scale_(
+                prediction_velocity,
+                prediction_pressure
+            )
+
+        return prediction_velocity.T, prediction_pressure.T
\ No newline at end of file