Estimated number of segments for 4 different penalized criteria in the regular case (top) and the irregular case (bottom). Top : Results of the simulations for 5 regularly spaced segments with n = 100 data points. The graph represents the average estimated number of segments for each criterion according to the standard deviation of the noise (σ). Bottom: Results of the simulations for 5 unregularly spaced segments with n = 100 data points. The adaptive criterion is robust to the additional noise since it maintains an estimate close to 5 segments whatever the noise and the configuration.
Picard et al. BMC Bioinformatics 2005 6:27 doi:10.1186/1471-2105-6-27