How to do … in ELKI

• Invoke ELKI from within Java
• Use index structures for accelerating data mining algorithms
• Process geo data in ELKI
• Perform geo-spatial outlier detection with an external neighborhood specification
• Precompute distance matrixes and use external distances
• Using sparse vector data
• more to come

See also the Tutorial section:

• Implementing a custom distance function, a variable exponent Minkowski-norm
• Implementing a new outlier detection algorithm, using the distances standard deviation
• Implementing a k-means clustering variant, producing clusters of the same size

and Examples:

• Greedy outlier ensemble computes a large set of outlier detection methods, then constructs and evaluates a greedy ensemble based on these methods.

• Cluster evaluation with segmented view detailes the configurations used for the ICDE 2012 demonstration on cluster evaluation.