subspace: Subspace: An R-Interface to the Subspace and Projected...

Description Details References


This package provides access to five of the most popular clustering algorithms in the subspace paradigm.


The algorithms CLIQUE, P3C, ProClus, SubClu and FIRES can be applied to data.frames and matrices and will return S3 objects representing clusterings. For example, using the built-in demo dataset, you can do:

>clustering <- P3C(subspace_dataset,PoissonThreshold=2)
Subspace clustering generated by the package Subspace, containing 12 clusters.

These subspace_clustering objects are actually just lists of subspace_cluster objects, which can be accessed as follows.

Subspace cluster generated by the package Subspace. This cluster consists of 140 objects in a 3 dimensional subspace.

Each of these clusters then holds a vector representing its subspace and a vector with the indexes of the objects the belong in this cluster. In this example, these could be accessed as clustering[[1]]$objects and clustering[[1]]$subspace.

This package also provides a plot method for subspace_clustering objects:

Showing dynamic visualisation. Press Escape/Ctrl + C to stop.

These plots are created using the ggvis package.

Finally, you can save clusterings to a file using the clustering_from_file and clustering_to_file functions.

For example you could save the clustering from this example to a file and load the true clustering of the demo dataset:

>path_to_clustering <- paste(path.package("subspace"),"/extdata/subspace_dataset.true",sep="")
true_clustering <- clustering_from_file(file_path=path_to_clustering)


M<c3><bc>ller E., G<c3><bc>nnemann S., Assent I., Seidl T.: Evaluating Clustering in Subspace Projections of High Dimensional Data In Proc. 35th International Conference on Very Large Data Bases (VLDB 2009), Lyon, France. (2009)

subspace documentation built on May 30, 2017, 2:39 a.m.

Related to subspace in subspace...