K-Means

Kmeans is one of the most popular clustering algorithm. It is an unsupervised method that can partition the data into k clusters. Kmeans is based on the Euclidean metric, but it can use other metrics. It provides an intuitive way of splitting data into groups.

Our implementation offers several important features:

• Fast kmeans implementation (for large k) based on trees
• Smart initialization of the centroids with the kmeans++ algorithm
• Automatic determination of optimal k based on BIC in batch mode and with the xmeans algorithm
• Cross validation

Related articles:

• K-Means
  • Introduction
  • A Simple Algorithm for a Fixed Number of Clusters
  • Cross-Validating for the Number of Clusters
  • Applications of K-Means
  • 1305 Implementation Details
  • Supported Features

Downloading

You can download it from here

Usage examples

• Running K-Means
• A Simple Example
• Kmeans with kd-tree and ball-tree
• Kmeans with sparse data
• Kmeans, trees versus naive mode
• Kmeans with online preprocessing
• Automatic determination of k
• Initialization of the centroids with kmeans++
• KMeans with automatic multiple restarts
• Finding the cluster memberships for unseen points

Suggestions and bug reports

We would like to know your opinion about our product. Please send us bug reports and suggestions at support@analytics1305.com

If you didn't find an answer to your question, check the forum http://www.analytics1305.com/forum