| Title: | Angle-Based Outlier Detection |
| Version: | 0.1 |
| Author: | Jose Jimenez <jose@jimenezluna.com> |
| Maintainer: | Jose Jimenez <jose@jimenezluna.com> |
| Description: | Performs angle-based outlier detection on a given dataframe. Three methods are available, a full but slow implementation using all the data that has cubic complexity, a fully randomized one which is way more efficient and another using k-nearest neighbours. These algorithms are specially well suited for high dimensional data outlier detection. |
| Depends: | cluster, R (≥ 3.1.2) |
| License: | MIT + file LICENSE |
| LazyData: | true |
| Encoding: | UTF-8 |
| NeedsCompilation: | no |
| Packaged: | 2015-08-30 21:39:13 UTC; hawk31 |
| Repository: | CRAN |
| Date/Publication: | 2015-08-31 14:31:42 |
Angle-Based Outlier Factor
Description
Computes angle-based outlier factor for each observation in the dataset
Usage
abod(data, method = "complete", n_sample_size = trunc(nrow(data)/10), k = 15)
Arguments
data |
Dataframe in which to compute angle-based outlier factor. |
method |
Method to perform. 'complete' will use the entire dataset (cubic complexity) to compute abof. 'randomized' will use a random sample of the data of size 'n_sample_size'. 'knn' will compute abof among 'k' nearest neighbours. |
n_sample_size |
Number of random observations to choose in randomized method. |
k |
Number of nearest neighbours to choose in knn method. |
Details
Please note that 'knn' has to compute an euclidean distance matrix before computing abof.
Value
Returns angle-based outlier factor for each observation. A small abof respect the others would indicate presence of an outlier.
Author(s)
Jose Jimenez <jose@jimenezluna.com>
References
[1] Angle-Based Outlier Detection in High-dimensional Data. KDD 2008. Hans-Peter Kriegel, Matthias Schubert, Arthur Zimek. (http://www.dbs.ifi.lmu.de/Publikationen/Papers/KDD2008.pdf)
Examples
abod(faithful, method = "randomized", n_sample_size = 5)
abod(faithful, method = "knn", k = 5)
Angle-based outlier detection
Description
Performs angle-based outlier detection on data. A complete, a randomized and a knn based methods are available.
| Package: | abodoutlier |
| Type: | Package |
| Version: | 0.1 |
| Date: | 2015-08-30 |
| License: | MIT License |
| Maintainer: | Jose Jimenez <jose@jimenezluna.com> |