not: Narrowest-Over-Threshold Change-Point Detection

Provides efficient implementation of the Narrowest-Over-Threshold methodology for detecting an unknown number of change-points occurring at unknown locations in one-dimensional data following 'deterministic signal + noise' model. Currently implemented scenarios are: piecewise-constant signal, piecewise-constant signal with a heavy-tailed noise, piecewise-linear signal, piecewise-quadratic signal, piecewise-constant signal and with piecewise-constant variance of the noise. For details, see Baranowski, Chen and Fryzlewicz (2019) <doi:10.1111/rssb.12322>.

install.packages("not")

Try the not package in your browser