A native R implementation of the PELT and CAPA algorithms for changepoint and anomaly detection.
The implementation is build around an S3 class anomalous_partition for storing inforamtion about the changepoints and anomalies and R6 classes for the cost functions relating to fitting different distributions to the data. The CAPA and PELT implementations make use of the methods of the anomalous_partition class. This allows for the replacement of the current anomalous_partition class with a new class with similarly names methods. The anomalous_partition class makes us of the cost classes via named methods, again allowing for generalisation and extension with further cost functions.
Using the code
This package is not available on CRAN. The latest development version can be installed from the git repository from within R using the devtools package:
devtools::install_github("waternumbers/anomalous")Prebuild packages of the development version available from the r-universe