R Package ‘gets‘:
Together with Genaro Sucarrat and James Reade I develop and maintain the R-package gets, which enables general-to-specific model selection and the detection of structural breaks and outliers through indicator saturation using impulses (IIS), step-indicators (SIS), and trend-indicators (TIS).
For more technical detail see the paper on SIS (open-access in Econometrics, 3:2, pp.240-264) and our paper on our R package (Journal of Statistical Software, 86(3)). We also have a blog post on using indicator saturation to detect breaks.
How many times the software package has been download (statistics from CRAN):
– Total so far:
– Last month:
The latest version of the package is version 0.16 (released on 28.8.2018, release-notes). Main updates in latest versions: parallel computing in isat to increase computational speed, and allowing for user-defined estimators. Please get in touch if there are any questions on the R-package: felix.pretis (at) nuffield.ox.ac.uk.
Try interactive indicator saturation using isat online (with thanks to James Reade for the shiny code):
Select a dataset (e.g. flow of the river Nile at Ashwan) and use indicator saturation to find sudden shifts (structural breaks) and outlying observations at a chosen level of significance (choose random colours if you’re feeling bold!).