Segmentation using Optimal Partitioning and Function Pruning

A dynamic programming algorithm for the fast segmentation of univariate signals into piecewise constant profiles. The 'fpop' package is a wrapper to a C++ implementation of the fpop (Functional Pruning Optimal Partioning) algorithm described in Maidstone et al. 2017 . The problem of detecting changepoints in an univariate sequence is formulated in terms of minimising the mean squared error over segmentations. The fpop algorithm exactly minimizes the mean squared error for a penalty linear in the number of changepoints.