message

client_1.py

@@ -0,0 +1,142 @@
+import flwr as fl
+import pandas as pd
+import numpy as np
+from sklearn.model_selection import train_test_split
+from tensorflow.keras.models import Sequential
+from tensorflow.keras.layers import Dense
+
+# Function to load your custom CSV data
+def load_custom_csv_data():
+    # Load your CSV data
+    # Replace 'path/to/your/data.csv' with the actual path to your CSV file
+    data = pd.read_csv('C:\\Users\\HP\\Desktop\\Trial-css\\heart_disease_data.csv')
+
+    # Split the data into features (X) and labels (y)
+    X = data.drop(columns=['target']).to_numpy()
+    y = data['target'].to_numpy()
+
+    return X, y
+
+# Define a simple model creation function
+def create_your_model(input_dim):
+    # Replace this with your own model creation code
+    # This is just a placeholder example
+    model = Sequential()
+    model.add(Dense(64, input_dim=input_dim, activation='relu'))
+    model.add(Dense(2, activation='softmax'))  # Assuming binary classification
+
+    # Compile the model with an appropriate loss and optimizer
+    model.compile(loss='sparse_categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
+
+    return model
+
+# Define a class that inherits from fl.client.NumPyClient
+from typing import Tuple, Dict
+
+class YourCustomClient(fl.client.NumPyClient):
+    def __init__(self, model, X_train, y_train, X_test, y_test, address):
+        self.model = model
+        self.X_train = X_train
+        self.y_train = y_train
+        self.X_test = X_test
+        self.y_test = y_test
+        self.address = address
+
+    def get_parameters(self, config):
+        return self.model.get_weights()
+
+    def predict(self, input_data):
+        """Make predictions using the trained model."""
+    # Ensure the input_data is in the same format as during training
+        input_data = np.asarray(input_data).astype(np.float32)
+
+    # Make predictions using the trained model
+        predictions = self.model.predict(input_data)
+
+    # Assuming binary classification with a threshold of 0.5
+        threshold = 0.5
+        binary_prediction = (predictions > threshold).astype(int)
+        return binary_prediction      
+
+    def set_parameters(self, parameters):
+        self.model.set_weights(parameters)
+
+    def fit(self, parameters, config):
+        print(f"Client {self.address} - Length of X_train: {len(self.X_train)}, Length of y_train: {len(self.y_train)}")
+        print(f"Client {self.address} - Setting parameters...")
+        self.set_parameters(parameters)
+        # Replace this with your actual training logic
+        print(f"Client {self.address} - Training on local data...")
+        history = self.model.fit(self.X_train, self.y_train, epochs=4)
+        # Get the number of examples processed during training
+        num_examples_processed = len(self.X_train)
+        # Return the updated model parameters, the number of examples processed, and additional metrics
+        updated_params = self.get_parameters(config=config)
+        metrics = {"accuracy": history.history.get("accuracy", [0])[-1]}  # Add your metrics here
+        print(f"Client {self.address} - Training complete. Updated model parameters:")
+        print(updated_params)
+        print(f"Number of examples processed: {num_examples_processed}")
+        print(f"Metrics: {metrics}")
+        # Make predictions on test data
+        test_predictions = self.predict(self.X_test)
+        threshold = 0.5
+        binary_test_prediction = (test_predictions[:, 1] > threshold).astype(int)
+        # Print or use the binary test prediction as needed
+        print(f"Client {self.address} - Binary Prediction on test data:")
+        print(binary_test_prediction)
+        return updated_params, num_examples_processed, metrics
+
+    def evaluate(self, parameters, config):
+        print(f"Client {self.address} - Length of X_test: {len(self.X_test)}, Length of y_test: {len(self.y_test)}")
+        print(f"Client {self.address} - Setting parameters for evaluation...")
+        self.set_parameters(parameters)
+        # Replace this with your actual evaluation logic
+        print(f"Client {self.address} - Evaluating on test data...")
+        evaluation_metrics = self.model.evaluate(self.X_test, self.y_test)
+        # Get the number of examples processed during evaluation
+        num_examples_processed = len(self.X_test)
+        # Return the evaluation metrics, the number of examples processed, and additional metrics (if any)
+        accuracy_position = 1  # Adjust accordingly based on your model's evaluate output
+        additional_metrics = {"accuracy": evaluation_metrics[accuracy_position]}
+        return evaluation_metrics[0], num_examples_processed, additional_metrics
+
+
+
+# Specify the server address
+server_address = "localhost:5000"
+
+# Load your custom CSV data
+# Load your custom CSV data
+X, y = load_custom_csv_data()
+
+# Print the shape of the entire dataset
+print("X shape:", X.shape)
+print("y shape:", y.shape)
+
+# Split the data into training and test sets
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
+
+# Print the shape of the training data
+print("X_train shape:", X_train.shape)
+print("y_train shape:", y_train.shape)
+
+
+# Create your model
+model = create_your_model(input_dim=X_train.shape[1])
+
+# Create an instance of YourCustomClient
+# Create an instance of YourCustomClient
+client = YourCustomClient(model, X_train, y_train, X_test, y_test, address="localhost:5000")
+
+# Print the length of the training data in the client
+# print(f"Client {client.address} - Length of X_train: {len(client.X)}, Length of y_train: {len(client.y)}")
+
+print(f"Connecting to Flower server at {server_address}...")
+print("Sending updated model to Flower server...")
+fl.client.start_client(
+    server_address=server_address,
+    client=client.to_client(),
+)
+print("Model sent successfully.")
+
+print("Connection successful. Exiting.")

client_2.py

@@ -0,0 +1,144 @@
+import flwr as fl
+import pandas as pd
+import numpy as np
+from sklearn.model_selection import train_test_split
+from tensorflow.keras.models import Sequential
+from tensorflow.keras.layers import Dense
+
+# Function to load your custom CSV data
+def load_custom_csv_data():
+    # Load your CSV data
+
+    # Replace 'path/to/your/data.csv' with the actual path to your CSV file
+    data = pd.read_csv('C:\\Users\\HP\\Desktop\\Trial-css\\heart_disease_data.csv')
+
+    # Split the data into features (X) and labels (y)
+    X = data.drop(columns=['target']).to_numpy()
+    y = data['target'].to_numpy()
+
+    return X, y
+
+# Define a simple model creation function
+def create_your_model(input_dim):
+    # Replace this with your own model creation code
+    # This is just a placeholder example
+    model = Sequential()
+    model.add(Dense(64, input_dim=input_dim, activation='relu'))
+    model.add(Dense(2, activation='softmax'))  # Assuming binary classification
+
+    # Compile the model with an appropriate loss and optimizer
+    model.compile(loss='sparse_categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
+
+    return model
+
+# Define a class that inherits from fl.client.NumPyClient
+from typing import Tuple, Dict
+
+class YourCustomClient(fl.client.NumPyClient):
+    def __init__(self, model, X_train, y_train, X_test, y_test, address):
+        self.model = model
+        self.X_train = X_train
+        self.y_train = y_train
+        self.X_test = X_test
+        self.y_test = y_test
+        self.address = address
+
+    def get_parameters(self, config):
+        return self.model.get_weights()
+
+    def predict(self, input_data):
+        """Make predictions using the trained model."""
+    # Ensure the input_data is in the same format as during training
+        input_data = np.asarray(input_data).astype(np.float32)
+
+    # Make predictions using the trained model
+        predictions = self.model.predict(input_data)
+
+    # Assuming binary classification with a threshold of 0.5
+        threshold = 0.5
+        binary_prediction = (predictions > threshold).astype(int)
+
+        return binary_prediction
+
+
+
+    def set_parameters(self, parameters):
+        self.model.set_weights(parameters)
+
+    def fit(self, parameters, config):
+        print(f"Client {self.address} - Length of X_train: {len(self.X_train)}, Length of y_train: {len(self.y_train)}")
+
+        print(f"Client {self.address} - Setting parameters...")
+        self.set_parameters(parameters)
+
+        # Replace this with your actual training logic
+        print(f"Client {self.address} - Training on local data...")
+        history = self.model.fit(self.X_train, self.y_train, epochs=4)
+
+        # Get the number of examples processed during training
+
+        num_examples_processed = len(self.X_train)
+
+
+
+        # Return the updated model parameters, the number of examples processed, and additional metrics
+        updated_params = self.get_parameters(config=config)
+        metrics = {"accuracy": history.history.get("accuracy", [0])[-1]}  # Add your metrics here
+        print(f"Client {self.address} - Training complete. Updated model parameters:")
+        print(updated_params)
+        print(f"Number of examples processed: {num_examples_processed}")
+        print(f"Metrics: {metrics}")
+
+        # Make predictions on test data
+        test_predictions = self.predict(self.X_test)
+
+        threshold = 0.5
+        binary_test_prediction = (test_predictions[:, 1] > threshold).astype(int)
+
+        # Print or use the binary test prediction as needed
+        print(f"Client {self.address} - Binary Prediction on test data:")
+        print(binary_test_prediction)
+
+        return updated_params, num_examples_processed, metrics
+
+
+
+
+# Specify the server address
+server_address = "localhost:5000"
+
+# Load your custom CSV data
+X, y = load_custom_csv_data()
+
+# Print the shape of the entire dataset
+print("X shape:", X.shape)
+print("y shape:", y.shape)
+
+# Split the data into training and test sets
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
+
+# Print the shape of the training data
+print("X_train shape:", X_train.shape)
+print("y_train shape:", y_train.shape)
+
+
+# Create your model
+model = create_your_model(input_dim=X_train.shape[1])
+
+
+# Create an instance of YourCustomClient
+client = YourCustomClient(model, X_train, y_train, X_test, y_test, address="localhost:5000")
+
+# Print the length of the training data in the client
+# print(f"Client {client.address} - Length of X_train: {len(client.X)}, Length of y_train: {len(client.y)}")
+
+# Start Flower client using start_client() and to_client()
+print(f"Connecting to Flower server at {server_address}...")
+print("Sending updated model to Flower server...")
+fl.client.start_client(
+    server_address=server_address,
+    client=client.to_client(),
+)
+print("Model sent successfully.")
+
+print("Connection successful. Exiting.")