KMeans Parallel Clustering Library

This library provides an implementation of the KMeans clustering algorithm with parallel processing using MPI for Python (mpi4py). It leverages multiprocessing and HDF5 for data serialization to enhance the efficiency and scalability of clustering large datasets.

Features

• Parallel KMeans Clustering: Distributes the clustering computation across multiple processors to accelerate the processing time.
• HDF5 Serialization: Uses HDF5 to handle large datasets efficiently by reading and writing data in a binary format.
• Easy Integration: Designed to be easily integrated into existing Python projects requiring KMeans clustering.

Usage

Initializing the KMeans Class

Create an instance of the KMeans class, specifying the number of clusters, maximum iterations, and optionally initial cluster centers.

from kmeans import KMeans
import numpy as np

# Example data
data = np.random.rand(100, 2)

# Initialize KMeans with 3 clusters
kmeans = KMeans(n_clusters=3,cluster_centers=[[1,2],[2,3],[3,4]])

# Fit the model to the data
kmeans.fit(data)

KMeans Class

The KMeans class handles the core functionality, including initialization, fitting the model to the data, and defining the method for parallel computation.

• init(self, n_clusters=3, max_iter=1000, cluster_centers=None): Initializes the KMeans object with the specified number of clusters, maximum iterations, and optional initial cluster centers.
• fit(self, data): Fits the KMeans model to the provided data, iterating up to the maximum number of iterations to adjust cluster centers and assign data points to clusters.
• predict(self): Placeholder method for future implementation to predict cluster membership for new data points.
• paralleleMethod(self): Handles the parallel computation of cluster assignments and updates cluster centers using MPI.

Parallel Computation with subProcess.py

The parallel computation is handled by a separate script, subProcess.py, executed by the MPI environment:

• MPI Initialization: Sets up the MPI environment, determines the rank and size of the processors.
• Data Distribution: Distributes data among available processors, ensuring an even workload.
• Cluster Assignment: Each processor calculates the distance of its data points to the cluster centers and assigns them to the nearest cluster.
• Data Gathering: Gathers the processed data from all processors and updates the cluster centers on the root processor.
• Cluster Center Update: Recalculates the cluster centers based on the new assignments and updates the HDF5 file with the new centers.
• Convergence Check: Checks if the cluster centers have stabilized; if so, terminates the computation.

Running the Code

1. Set up the environment: Ensure you have the necessary dependencies installed.
2. Create an instance of KMeans: Initialize with the desired number of clusters and fit the model to your data.
3. Run the parallel processing script: Ensure subProcess.py is in the same directory and is executed correctly by MPI.

Replace <number_of_processes> with the number of processors you want to use for parallel computation.

Note

• This implementation is a basic example of parallelizing KMeans clustering. Further optimizations and improvements can be made for more robust and efficient performance.
• The predict method is not implemented yet and will need to be developed to use the trained model for making predictions on new data.

License

This project is licensed under the MIT License - see the LICENSE file for details. Acknowledgements

• This implementation uses numpy for numerical operations.
• HDF5 file handling is done using the h5py library.
• Parallel processing is facilitated by the mpi4py library.

Contact

For any questions or contributions, feel free to open an issue or submit a pull request on the project's GitHub repository.