TY - JOUR
T1 - Segmentation of circular data
AU - Hawkins, Douglas M.
AU - Lombard, F.
PY - 2015/1/31
Y1 - 2015/1/31
N2 - Circular data – data whose values lie in the interval [0,2π) – are important in a number of application areas. In some, there is a suspicion that a sequence of circular readings may contain two or more segments following different models. An analysis may then seek to decide whether there are multiple segments, and if so, to estimate the changepoints separating them. This paper presents an optimal method for segmenting sequences of data following the von Mises distribution. It is shown by example that the method is also successful in data following a distribution with much heavier tails.
AB - Circular data – data whose values lie in the interval [0,2π) – are important in a number of application areas. In some, there is a suspicion that a sequence of circular readings may contain two or more segments following different models. An analysis may then seek to decide whether there are multiple segments, and if so, to estimate the changepoints separating them. This paper presents an optimal method for segmenting sequences of data following the von Mises distribution. It is shown by example that the method is also successful in data following a distribution with much heavier tails.
KW - changepoints
KW - model selection
KW - von Mises distribution
UR - http://www.scopus.com/inward/record.url?scp=84988273872&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84988273872&partnerID=8YFLogxK
U2 - 10.1080/02664763.2014.934665
DO - 10.1080/02664763.2014.934665
M3 - Article
AN - SCOPUS:84988273872
SN - 0266-4763
VL - 42
SP - 88
EP - 97
JO - Journal of Applied Statistics
JF - Journal of Applied Statistics
IS - 1
ER -