def analyze_pmp(ts, query, sample_pct, threshold, windows=None, n_jobs=-1):
"""
Computes the Pan-MatrixProfile, top 3 motifs and top 3 discords for the
provided time series and query. Additionally, plots for the PMP, motifs
and discords is provided.
Parameters
----------
ts : array_like
The time series to analyze.
query : array_like
The query to analyze.
sample_pct : float
A float between 0 and 1 representing how many samples to compute for
the PMP.
threshold : float
A correlation threshold between 0 and 1 that is used to compute the
upper window. Note that this is used only when the windows is None.
windows : array_like, default None
Integers representing the desired windows to use during the
computation of the PMP.
n_jobs : int, default -1 (all cpu cores)
The number of cpu cores to use when computing the PMP.
Returns
-------
tuple : (profile, figures)
A tuple with the first item being the profile and the second being an
array of matplotlib figures.
"""
ts = core.to_np_array(ts)
if isinstance(threshold, type(None)):
threshold = 0.98
# when a threshold is passed, we compute the upper window
profile = None
if isinstance(windows, type(None)):
profile = maximum_subsequence(ts, threshold, include_pmp=True)
# determine windows to be computed
# from 8 in steps of 2 until upper w
start = 8
windows = range(start, profile['upper_window'] + 1)
# compute the pmp
profile = skimp(ts, windows=windows, sample_pct=sample_pct,
pmp_obj=profile)
# extract top motifs
profile = motifs(profile)
# extract top discords
profile = discords(profile)
# plot pmp
figures = visualize(profile)
return (profile, figures)
def analyze_mp_approximate(ts, query, window, sample_pct, n_jobs=-1):
"""
Computes the exact MatrixProfile, top 3 motifs and top 3 discords for the
provided time series and query. Additionally, the MatrixProfile, discords
and motifs are visualized.
Parameters
----------
ts : array_like
The time series to analyze.
query : array_like
The query to analyze.
window : int
The window size to compute the MatrixProfile.
sample_pct : float
A float between 0 and 1 representing how many samples to compute for
the MP. When it is 1, it is the same as using the exact algorithm.
n_jobs : int, default -1 (all cpu cores)
The number of cpu cores to use when computing the MP.
Returns
-------
tuple : (profile, figures)
A tuple with the first item being the profile and the second being an
array of matplotlib figures.
"""
ts = core.to_np_array(ts)
# compute mp
profile = scrimp_plus_plus(ts,
window,
query=query,
sample_pct=sample_pct,
n_jobs=n_jobs)
# extract top motifs
profile = motifs(profile)
# extract top discords
profile = discords(profile)
# plot mp
figures = visualize(profile)
return (profile, figures)
def analyze_mp_exact(ts, query, window, n_jobs=-1):
"""
Computes the exact MatrixProfile, top 3 motifs and top 3 discords for the
provided time series and query. Additionally, the MatrixProfile, discords
and motifs are visualized.
Parameters
----------
ts : array_like
The time series to analyze.
query : array_like
The query to analyze.
window : int
The window size to compute the MatrixProfile.
n_jobs : int, default -1 (all cpu cores)
The number of cpu cores to use when computing the MP.
Returns
-------
tuple : (profile, figures)
A tuple with the first item being the profile and the second being an
array of matplotlib figures.
"""
ts = core.to_np_array(ts)
# compute mp
profile = mpx(ts, window, query=query, n_jobs=n_jobs)
# extract top motifs
profile = motifs(profile)
# extract top discords
profile = discords(profile)
# plot mp
figures = visualize(profile)
return (profile, figures)