Wraps CptNonPar::np.mojo() (McGonigle and Cho, 2025): nonparametric
moving-sum detection of changes in the marginal or joint distribution of a
(possibly multivariate) time series, robust to serial dependence.
Arguments
- x
A numeric vector or matrix (rows are time points).
- G
Moving-window bandwidth. Defaults to
max(20, 0.1 * n)observations, capped atn / 2— the largest bandwidth the engine accepts — so the default also works on series shorter than 40.- lag
Time lag at which changes in the joint distribution are examined;
0targets the marginal distribution. Defaults to0.- ...
Additional arguments passed to
CptNonPar::np.mojo().
Value
A ggcpt object. Constant coordinates leave the kernel
statistics undefined, so they are dropped (with a warning) before
detection and an all-constant input returns an empty result. The
engine calibrates its detection threshold by bootstrap, so the value
recorded in the penalty descriptor varies between runs; call
set.seed() beforehand, or pass threshold = "manual" and
threshold.val through ..., for a reproducible one.
References
McGonigle ET, Cho H (2025). “Nonparametric data segmentation in multivariate time series via joint characteristic functions.” Biometrika, 112(2), asaf024.
See also
Other changepoint engines:
bcp_wrapper(),
beast_wrapper(),
bfast_wrapper(),
binsegrcpp_wrapper(),
bocpd_wrapper(),
cpm_wrapper(),
cpop_wrapper(),
cpt_wrapper(),
decafs_wrapper(),
ecp_wrapper(),
envcpt_wrapper(),
esac_wrapper(),
fabisearch_wrapper(),
fastcpd_wrapper(),
fcov_wrapper(),
fmean_wrapper(),
fpop_wrapper(),
geomcp_wrapper(),
hdcov_wrapper(),
hdreg_wrapper(),
idetect_wrapper(),
inspect_wrapper(),
kcp_wrapper(),
kwc_wrapper(),
mcp_wrapper(),
mosum_wrapper(),
network_wrapper(),
not_wrapper(),
nsp_wrapper(),
ocd_wrapper(),
pilliat_wrapper(),
segmented_wrapper(),
smuce_wrapper(),
sn_wrapper(),
strucchange_wrapper(),
taylor_wrapper(),
tguh_wrapper(),
trend_wrapper(),
var_wrapper(),
wbs2_wrapper(),
wbs_wrapper(),
wbsts_wrapper()