Add examples for all helpers functions

examples/helpers/__init__.py

@@ -0,0 +1,5 @@
+#!/usr/bin/env python
+# Created by "Thieu" at 22:28, 14/08/2023 ----------%
+#       Email: nguyenthieu2102@gmail.com            %                                                    
+#       Github: https://github.com/thieu1995        %                         
+# --------------------------------------------------%

examples/helpers/exam_encoder.py

@@ -0,0 +1,21 @@
+#!/usr/bin/env python
+# Created by "Thieu" at 23:29, 24/09/2023 ----------%                                                                               
+#       Email: nguyenthieu2102@gmail.com            %                                                    
+#       Github: https://github.com/thieu1995        %                         
+# --------------------------------------------------%
+
+import numpy as np
+from metaperceptron import Data
+
+X = np.array([[1., -2., 2.],
+              [-2., 1., 3.],
+              [4., 1., -2.]])
+y = np.array([[1, 2, 0],
+              [0, 0, 1],
+              [0, 2, 2]])
+
+y = np.array([[1, 2, 0]])
+
+data = Data(X, y)
+y, le = data.encode_label(y)
+print(y)

examples/helpers/exam_get_weights.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python
+# Created by "Thieu" at 15:18, 17/09/2023 ----------%                                                                               
+#       Email: nguyenthieu2102@gmail.com            %                                                    
+#       Github: https://github.com/thieu1995        %                         
+# --------------------------------------------------%
+
+from metaperceptron import MhaMlpClassifier, Data
+from sklearn.datasets import make_classification
+
+
+# Create a multi-class classification dataset with 4 classes
+X, y = make_classification(
+    n_samples=300,  # Total number of data points
+    n_features=7,  # Number of features
+    n_informative=3,  # Number of informative features
+    n_redundant=0,  # Number of redundant features
+    n_classes=4,  # Number of classes
+    random_state=42
+)
+data = Data(X, y, name="RandomData")
+data.split_train_test(test_size=0.2, random_state=2)
+
+opt_paras = {"name": "WOA", "epoch": 10, "pop_size": 30}
+model = MhaMlpClassifier(hidden_size=50, act1_name="tanh", act2_name="sigmoid",
+                         obj_name="NPV", optimizer="OriginalWOA", optimizer_paras=opt_paras, verbose=True)
+model.fit(data.X_train, data.y_train)
+y_pred = model.predict(data.X_test)
+
+## Get parameters for model
+print(model.get_params())
+
+## Get weights of neural network (ELM network)
+print(model.network.get_weights())

examples/helpers/exam_lb_ub.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python
+# Created by "Thieu" at 09:49, 25/09/2023 ----------%                                                                               
+#       Email: nguyenthieu2102@gmail.com            %                                                    
+#       Github: https://github.com/thieu1995        %                         
+# --------------------------------------------------%
+
+from metaperceptron import MhaMlpClassifier, Data
+from sklearn.datasets import make_classification
+
+
+# Create a multi-class classification dataset with 4 classes
+X, y = make_classification(
+    n_samples=300,  # Total number of data points
+    n_features=7,  # Number of features
+    n_informative=3,  # Number of informative features
+    n_redundant=0,  # Number of redundant features
+    n_classes=4,  # Number of classes
+    random_state=42
+)
+data = Data(X, y, name="RandomData")
+data.split_train_test(test_size=0.2, random_state=2)
+
+opt_paras = {"name": "WOA", "epoch": 30, "pop_size": 30}
+model = MhaMlpClassifier(hidden_size=50, act1_name="tanh", act2_name="sigmoid",
+                 obj_name="CEL", optimizer="OriginalWOA", optimizer_paras=opt_paras, verbose=True)
+model.fit(data.X_train, data.y_train, lb=(-10., ), ub=(10., ))
+y_pred = model.predict(data.X_test)
+
+## Get parameters for model
+print(model.get_params())
+
+## Get weights of neural network (ELM network)
+print(model.network.get_weights())

examples/helpers/exam_save_load_model.py

@@ -0,0 +1,38 @@
+#!/usr/bin/env python
+# Created by "Thieu" at 15:10, 17/09/2023 ----------%                                                                               
+#       Email: nguyenthieu2102@gmail.com            %                                                    
+#       Github: https://github.com/thieu1995        %                         
+# --------------------------------------------------%
+
+from metaperceptron import MhaMlpClassifier, Data
+from sklearn.datasets import make_classification
+
+
+# Create a multi-class classification dataset with 4 classes
+X, y = make_classification(
+    n_samples=300,  # Total number of data points
+    n_features=7,  # Number of features
+    n_informative=3,  # Number of informative features
+    n_redundant=0,  # Number of redundant features
+    n_classes=4,  # Number of classes
+    random_state=42
+)
+data = Data(X, y, name="RandomData")
+data.split_train_test(test_size=0.2, random_state=2)
+
+opt_paras = {"name": "WOA", "epoch": 10, "pop_size": 30}
+model = MhaMlpClassifier(hidden_size=50, act1_name="tanh", act2_name="sigmoid",
+                 obj_name="CEL", optimizer="OriginalWOA", optimizer_paras=opt_paras, verbose=True)
+model.fit(data.X_train, data.y_train)
+y_pred = model.predict(data.X_test)
+print(model.evaluate(data.y_test, y_pred, list_metrics=("AS", "PS", "F1S")))
+model.save_training_loss(save_path="history", filename="loss.csv")
+model.save_evaluation_metrics(data.y_test, y_pred, list_metrics=("AS", "PS", "F1S"), save_path="history", filename="metrics.csv")
+
+## Save Model
+model.save_model(save_path="history", filename="ga-elm.pkl")
+
+## Load Model
+new_model = MhaMlpClassifier()
+trained_model = new_model.load_model(load_path="history", filename="ga-elm.pkl")
+print(trained_model.scores(data.X_test, data.y_test, list_methods=("AS", "PS", "F1S")))

examples/helpers/exam_save_results.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python
+# Created by "Thieu" at 13:52, 17/09/2023 ----------%                                                                               
+#       Email: nguyenthieu2102@gmail.com            %                                                    
+#       Github: https://github.com/thieu1995        %                         
+# --------------------------------------------------%
+
+from metaperceptron import MhaMlpClassifier, Data
+from sklearn.datasets import make_classification
+
+
+# Create a multi-class classification dataset with 4 classes
+X, y = make_classification(
+    n_samples=300,  # Total number of data points
+    n_features=7,  # Number of features
+    n_informative=3,  # Number of informative features
+    n_redundant=0,  # Number of redundant features
+    n_classes=4,  # Number of classes
+    random_state=42
+)
+data = Data(X, y, name="RandomData")
+data.split_train_test(test_size=0.2, random_state=2)
+
+opt_paras = {"name": "WOA", "epoch": 10, "pop_size": 30}
+model = MhaMlpClassifier(hidden_size=50, act1_name="tanh", act2_name="sigmoid",
+                         obj_name="CEL", optimizer="OriginalWOA", optimizer_paras=opt_paras, verbose=True)
+model.fit(data.X_train, data.y_train)
+y_pred = model.predict(data.X_test)
+print(model.evaluate(data.y_test, y_pred, list_metrics=("AS", "PS", "F1S")))
+
+## Save results
+model.save_training_loss(save_path="history", filename="loss.csv")
+
+model.save_evaluation_metrics(data.y_test, y_pred, list_metrics=("AS", "PS", "F1S"), save_path="history", filename="metrics.csv")

examples/helpers/exam_save_y_predicted.py

@@ -0,0 +1,32 @@
+#!/usr/bin/env python
+# Created by "Thieu" at 15:43, 17/09/2023 ----------%                                                                               
+#       Email: nguyenthieu2102@gmail.com            %                                                    
+#       Github: https://github.com/thieu1995        %                         
+# --------------------------------------------------%
+
+from metaperceptron import MhaMlpClassifier, Data
+from sklearn.datasets import make_classification
+
+
+# Create a multi-class classification dataset with 4 classes
+X, y = make_classification(
+    n_samples=300,  # Total number of data points
+    n_features=7,  # Number of features
+    n_informative=3,  # Number of informative features
+    n_redundant=0,  # Number of redundant features
+    n_classes=4,  # Number of classes
+    random_state=42
+)
+data = Data(X, y, name="RandomData")
+data.split_train_test(test_size=0.2, random_state=2)
+
+opt_paras = {"name": "WOA", "epoch": 10, "pop_size": 30}
+model = MhaMlpClassifier(hidden_size=50, act1_name="tanh", act2_name="sigmoid",
+                 obj_name="CEL", optimizer="OriginalWOA", optimizer_paras=opt_paras, verbose=True)
+model.fit(data.X_train, data.y_train)
+y_pred = model.predict(data.X_test)
+print(model.evaluate(data.y_test, y_pred, list_metrics=("AS", "PS", "F1S")))
+
+## Save predicted results
+model.save_y_predicted(data.X_train, data.y_train, save_path="history", filename="train_y_predicted.csv")
+model.save_y_predicted(data.X_test, data.y_test, save_path="history", filename="test_y_predicted.csv")

examples/helpers/exam_scaler.py

@@ -0,0 +1,81 @@
+#!/usr/bin/env python
+# Created by "Thieu" at 12:49, 17/09/2023 ----------%                                                                               
+#       Email: nguyenthieu2102@gmail.com            %                                                    
+#       Github: https://github.com/thieu1995        %                         
+# --------------------------------------------------%
+
+import numpy as np
+from metaperceptron import DataTransformer, Data
+
+
+X = np.array([[1., -2., 2.],
+              [-2., 1., 3.],
+              [4., 1., -2.]])
+
+## Get all supported scaling methods
+print(DataTransformer.SUPPORTED_SCALERS.keys())
+
+### 1) Using only standard scaler
+scaler = DataTransformer(scaling_methods="standard")
+X_scaled = scaler.fit_transform(X)
+X_unscaled = scaler.inverse_transform(X_scaled)
+
+# Print the results
+print("Original Data:")
+print(X)
+print("Scaled Data:")
+print(X_scaled)
+print("Transformed Back to Original:")
+print(X_unscaled)
+
+### 2) Using multiple  scalers
+scaler = DataTransformer(scaling_methods=("standard", "minmax"))  # Just like Pipeline
+X_scaled = scaler.fit_transform(X)
+X_unscaled = scaler.inverse_transform(X_scaled)
+
+# Print the results
+print("\nOriginal Data:")
+print(X)
+print("Scaled Data:")
+print(X_scaled)
+print("Transformed Back to Original:")
+print(X_unscaled)
+
+### 3) Use methods in Data class instead
+data = Data(X)
+X_scaled, scaler = data.scale(X, scaling_methods=("standard", "minmax"))  # Just like Pipeline
+X_unscaled = scaler.inverse_transform(X_scaled)
+
+# Print the results
+print("\nOriginal Data:")
+print(X)
+print("Scaled Data:")
+print(X_scaled)
+print("Transformed Back to Original:")
+print(X_unscaled)
+
+### 4) Use methods in Data class with parameters
+data = Data(X)
+X_scaled, scaler = data.scale(X, scaling_methods=("sinh-arc-sinh", "minmax"), list_dict_paras=({"epsilon": 0.5, "delta": 2.5}, None))
+X_unscaled = scaler.inverse_transform(X_scaled)
+
+# Print the results
+print("\nOriginal Data:")
+print(X)
+print("Scaled Data:")
+print(X_scaled)
+print("Transformed Back to Original:")
+print(X_unscaled)
+
+### 5) Use methods in Data class with parameters
+data = Data(X)
+X_scaled, scaler = data.scale(X, scaling_methods=("yeo-johnson", "sinh-arc-sinh"), list_dict_paras=({"lmbda": 1.2}, {"epsilon": 0.5, "delta": 2.5}))
+X_unscaled = scaler.inverse_transform(X_scaled)
+
+# Print the results
+print("\nOriginal Data:")
+print(X)
+print("Scaled Data:")
+print(X_scaled)
+print("Transformed Back to Original:")
+print(X_unscaled)