This method calculates the irregularity of a data point based on the positions of its multiple neighbours.
The timeline is segmented into two contrasting movement phases: 'move-around'- and 'move-away' phases.