ELKI stands for Environment for Loping KDD applications Index structures, where KDD stands for Knowledge Discovery in Database. It is an open source software used mainly for data mining, with an emphasis on unsupervised learning. It supports various algorithms for cluster analysis and outlier detection. The following are some outlier algorithms:
- Distance-based outlier detection: This is used to specify two parameters. The object is flagged outlier if its fraction, p, for all the data objects that have a distance above d from c. There are many algorithms, such as DBOutlierDetection, DBOutlierScore, KNNOutlier, KNNWeightOutlier, ParallelKNNOutlier, ParallelKNNWeightOutlier, ReferenceBasedOutlierDetection, and so on.
- LOF family methods: This computes density-based local outlier factors on specific parameters. It includes algorithms such as LOF, ParallelLOF...
