Cluster Branch - Implementation of manifold models to reduce dimensionality on data for visualization

bibmon/__init__.py

@@ -3,12 +3,13 @@
 from ._esn import ESN
 from ._sbm import SBM
 from ._sklearn_regressor import sklearnRegressor
+from ._sklearn_manifold import sklearnManifold
 from ._preprocess import PreProcess
 from ._load_data import load_tennessee_eastman, load_real_data
 from ._bibmon_tools import train_val_test_split, complete_analysis, comparative_table, spearmanr_dendrogram, create_df_with_dates, create_df_with_noise, align_dfs_by_rows
 
 __all__ = ['Autoencoder','PCA','ESN','SBM',
-	   'sklearnRegressor', 'PreProcess',
+	   'sklearnRegressor','sklearnManifold' , 'PreProcess',
            'load_tennessee_eastman', 'load_real_data', 
            'train_val_test_split', 'complete_analysis', 'comparative_table',
 	       'spearmanr_dendrogram', 'create_df_with_dates',

bibmon/_sklearn_manifold.py

@@ -0,0 +1,156 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Tue  8 08:14:01 2024
+
+@author: leovo
+"""
+
+
+import matplotlib.pyplot as plt
+import pandas as pd
+
+from ._generic_model import GenericModel
+
+###############################################################################
+
+class sklearnManifold(GenericModel):
+    """
+    Interface for sklearn manifold learning models.
+
+    Parameters
+    ----------
+    manifold_model: any manifold model that uses the sklearn interface. 
+        For example:
+            * sklearn.manifold.MDS,
+            * sklearn.manifold.Isomap,
+            * sklearn.manifold.TSNE,
+            * sklearn.manifold.LocallyLinearEmbedding,
+            * etc....
+    """
+
+    ###########################################################################
+
+    def __init__(self, manifold_model):
+        self.has_Y = False  # Default set to False, because Manifold algorithms don't require a target variable
+        self.manifold_model = manifold_model
+
+        self.name = self.manifold_model.__class__.__name__
+
+    ###########################################################################
+
+    def train_core(self):
+        """
+        Fits the manifold model using the training data.
+        """
+        ## Check if the input is a pandas DataFrame
+        if isinstance(self.X_train, pd.DataFrame):
+            # If it's a DataFrame, use the `.values` attribute to extract numpy array
+            self.transformed_data = self.manifold_model.fit_transform(self.X_train.values)
+        else:
+            # If it's already a numpy array, use it directly
+            self.transformed_data = self.manifold_model.fit_transform(self.X_train)
+
+        ###########################################################################
+
+    def fit_transform(self,X):
+        """
+        Fits the clustering method and returns the transformed data
+
+        """
+
+        self.X_train=X #Attributing training data to variable X passed in the m
+        self.train_core() #Training the method with train_core
+
+        """ 
+        Returning the transformed data for visualization
+        """
+        return self.transformed_data
+
+
+    def map_from_X(self,X_test):
+        """
+        Applies the transformation to a new dataset. Note that some manifold
+        models, like TSNE, may not have a direct `transform` method.
+        """
+        if hasattr(self.manifold_model, 'transform'):
+            return self.manifold_model.transform(X_test)
+        else:
+            raise NotImplementedError("This manifold model does not support transformation on new data.")
+
+    ###########################################################################
+
+    def set_hyperparameters(self, params_dict):
+        """
+        Sets the hyperparameters for the manifold model.
+        """
+        for key, value in params_dict.items():
+            setattr(self.manifold_model, key, value)
+
+    ###########################################################################
+
+    def transform(self, X_test):
+        """
+        Transforms the input data using the trained manifold model by calling map_from_X.
+
+        Parameters
+        ----------
+        X_test: array-like or DataFrame
+            The new data to transform.
+
+        Returns
+        -------
+        transformed_data: array-like
+            The transformed data.
+        """
+        return self.map_from_X(X_test)
+
+    def plot_embedding(self):
+        """
+        Plots the 2D or 3D embedding resulting from the manifold model.
+        """
+        if self.transformed_data.shape[1] == 2:
+            plt.scatter(self.transformed_data[:, 0], self.transformed_data[:, 1], s=50, cmap='viridis')
+            plt.title(f"{self.name} 2D Embedding")
+            plt.xlabel("Component 1")
+            plt.ylabel("Component 2")
+        elif self.transformed_data.shape[1] == 3:
+            fig = plt.figure()
+            ax = fig.add_subplot(111, projection='3d')
+            ax.scatter(self.transformed_data[:, 0], self.transformed_data[:, 1], self.transformed_data[:, 2], s=50, cmap='viridis')
+            ax.set_title(f"{self.name} 3D Embedding")
+            ax.set_xlabel("Component 1")
+            ax.set_ylabel("Component 2")
+            ax.set_zlabel("Component 3")
+        else:
+            print("Embedding dimensionality is not 2D or 3D; custom plotting is required.")
+
+    def clusters_visualization(self, X):
+        """
+        Fits the manifold model, transforms the data, and plots the resulting 2D or 3D embedding.
+
+        Parameters
+        ----------
+        X: array-like or DataFrame
+            The data to fit and transform.
+        """
+        # Perform fit_transform and store the transformed data
+        transformed_data = self.fit_transform(X)
+
+        # Plot the 2D or 3D embedding based on the transformed data
+        if transformed_data.shape[1] == 2:
+            plt.scatter(transformed_data[:, 0], transformed_data[:, 1], s=50, cmap='viridis')
+            plt.title(f"{self.name} 2D Embedding")
+            plt.xlabel("Component 1")
+            plt.ylabel("Component 2")
+        elif transformed_data.shape[1] == 3:
+            fig = plt.figure()
+            ax = fig.add_subplot(111, projection='3d')
+            ax.scatter(transformed_data[:, 0], transformed_data[:, 1], transformed_data[:, 2], s=50, cmap='viridis')
+            ax.set_title(f"{self.name} 3D Embedding")
+            ax.set_xlabel("Component 1")
+            ax.set_ylabel("Component 2")
+            ax.set_zlabel("Component 3")
+        else:
+            print("Embedding dimensionality is not 2D or 3D; custom plotting is required.")
+
+        plt.show()

test/test_manifold_models.py

@@ -0,0 +1,78 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Fri Nov  8 09:34:01 2024
+
+@author: Leonardo Voltolini
+"""
+
+import bibmon
+from sklearn.preprocessing import StandardScaler
+import numpy as np
+
+
+
+SC=StandardScaler()
+
+# loading the data from TEP
+df_train, df_test = bibmon.load_tennessee_eastman(train_id = 0, 
+                                                   test_id = 1)
+
+#Transforming training and testing data using StandardScaler
+X_train=SC.fit_transform(df_train)
+X_test=SC.transform(df_test)
+
+#Concatenating train and the test, because manifold models normally
+#don't require a separation between training and testing folds
+X=np.concatenate( (X_train, X_test),axis=0)
+
+
+for attr in bibmon.__all__:             
+    a = getattr(bibmon,attr)     
+    if isinstance(a, type):         
+        '''
+        Verifying if the attribute a is generic model from sklearn manifold
+        and then applying the adequate model as wanted
+        '''
+        if a.__base__ == bibmon._generic_model.GenericModel:   
+            if a == bibmon.sklearnManifold:             
+                from sklearn.manifold import TSNE
+                model = a(TSNE(n_components=2)) #Creating the model
+
+                '''
+                Computing the embeeding data from fit_transform function
+                and subsequently plotting the clustering in the appropriate 
+                dimension
+                '''
+                embedded_data=model.fit_transform(X)
+                model.plot_embedding()
+
+#%%
+
+'''
+This implementation does the same as previous cell, but, it applies
+a distinct model and automatically computes fit_transform and clusters 
+visualization
+'''
+
+for attr in bibmon.__all__:             
+    a = getattr(bibmon,attr)     
+    if isinstance(a, type):         
+        if a.__base__ == bibmon._generic_model.GenericModel:   
+            if a == bibmon.sklearnManifold:                 
+                from sklearn.manifold import MDS
+                model = a(MDS(n_components=3))
+
+                '''
+                The below code transforms the data and presents the
+                graph for cluster visualization
+                '''
+                model.clusters_visualization(X)
+
+
+
+
+
+
+
+
+