connorbrubaker/rulsif.ts: Time Series Change Point Detection Using Relative Unconstrained Least Squares Importance Fitting (RULSIF)

rulsif.ts implements relative unconstrained least squares importance fitting for the task of detecting change points in time series data. The main function is ts_detect which automatically creates a matrix of subsequences of the series and associates to each time point a "score" from which change points may be detected. The functions in comp_dist.R, comp_median.R, gaussian_kernel.R, RelULSIF.R, and sigma_lambda_grid_search.R were originally written in MATLAB by Song Liu.

install.packages("remotes")
remotes::install_github("connorbrubaker/rulsif.ts")