Page MenuHomec4science
Files F63510641

CUSUM.py
No OneTemporary
Actions

File Metadata

Created
Mon, May 20, 15:40

CUSUM.py
View Options

import numpy as np
import pyqtgraph as pg
from timeit import default_timer as timer
def detection(self, data, dt, threshhold , minlength , maxstates):
s = timer()
logp = 0 #instantaneous log-likelihood for positive jumps
logn = 0 #instantaneous log-likelihood for negative jumps
cpos = np.zeros(len(data), dtype='float64') #cumulative log-likelihood function for positive jumps
cneg = np.zeros(len(data), dtype='float64') #cumulative log-likelihood function for negative jumps
gpos = np.zeros(2, dtype='float64') #decision function for positive jumps
gneg = np.zeros(2, dtype='float64') #decision function for negative jumps
edges = np.array([0], dtype='int64') #initialize an array with the position of the first subevent - the start of the event
real_start = np.array([], dtype='int64') #initialize an array with the position of the first subevent - the start of the event
real_end = np.array([], dtype='int64') #initialize an array with the position of the first subevent - the start of the event
real_Depth = np.array([], dtype='int64') #initialize an array with the position of the first subevent - the start of the event
anchor = 0 #the last detected change
length = len(data)
self.var = np.std(data)
h = threshhold / self.var
k = 1000
nStates = 0
varM = data[0]
varS = 0
mean = data[0]
print('length data =' + str(length))
v = np.zeros(length, dtype='float64')
while k < length-100:
k += 1
if nStates == 0:
variance = np.var(data[anchor:k]) # initial params for pattern region
mean = np.mean(data[anchor:k])
if variance == 0: break
logp = threshhold/variance * (data[k] - mean - threshhold/2.) #instantaneous log-likelihood for current sample assuming local baseline has jumped in the positive direction
logn = -threshhold/variance * (data[k] - mean + threshhold/2.) #instantaneous log-likelihood for current sample assuming local baseline has jumped in the negative direction
cpos[k] = cpos[k-1] + logp #accumulate positive log-likelihoods
cneg[k] = cneg[k-1] + logn #accumulate negative log-likelihoods
gpos[1] = max(gpos[0] + logp, 0) #accumulate or reset positive decision function
gneg[1] = max(gneg[0] + logn, 0) #accumulate or reset negative decision function
if (gpos[1] > h or gneg[1] > h):
if (gpos[1] > h): #significant positive jump detected
jump = anchor + np.argmin(cpos[anchor:k+1]) #find the location of the start of the jump
if jump - edges[nStates] > minlength and np.abs(data[jump+minlength]-data[jump]) >threshhold/4:
edges = np.append(edges, jump)
nStates += 1
#print('EVENT!!!!! at ='+str(self.t[jump]))
anchor = k# no data meaning at bad points!
# away from bad point more!
cpos[0:len(cpos)] = 0 #reset all decision arrays
cneg[0:len(cneg)] = 0
gpos[0:len(gpos)] = 0
gneg[0:len(gneg)] = 0
if (gneg[1] > h): #significant negative jump detected
jump = anchor + np.argmin(cneg[anchor:k+1])
if jump - edges[nStates] > minlength and np.abs(data[jump+minlength]-data[jump]) >threshhold/4:
edges = np.append(edges, jump)
nStates += 1
#print('EVENT!!!!! at ='+str(self.t[jump] ))
anchor = k # no data meaning at bad points!
# away from bad point more!
cpos[0:len(cpos)] = 0 #reset all decision arrays
cneg[0:len(cneg)] = 0
gpos[0:len(gpos)] = 0
gneg[0:len(gneg)] = 0
gpos[0] = gpos[1]
gneg[0] = gneg[1]
if maxstates > 0:
if nStates > maxstates:
print('too sensitive')
nStates = 0
k = 0
threshhold = threshhold*1.1
h = h*1.1
logp = 0 #instantaneous log-likelihood for positive jumps
logn = 0 #instantaneous log-likelihood for negative jumps
cpos = np.zeros(len(data), dtype='float64') #cumulative log-likelihood function for positive jumps
cneg = np.zeros(len(data), dtype='float64') #cumulative log-likelihood function for negative jumps
gpos = np.zeros(2, dtype='float64') #decision function for positive jumps
gneg = np.zeros(2, dtype='float64') #decision function for negative jumps
edges = np.array([0], dtype='int64') #initialize an array with the position of the first subevent - the start of the event
anchor = 0 #the last detected change
length = len(data)
mean = data[0]
nStates = 0
mean = data[0]
edges = np.append(edges, len(data)-1) #mark the end of the event as an edge
nStates += 1
cusum = dict()
print('Events = ' + str(self.t[edges]))
for i in range(len(edges)-1):
if edges[i+1] - edges[i] < int(0.05 * self.outputsamplerate):
real_start = np.append(real_start, edges[i])
real_end = np.append(real_end, edges[i+1])
real_Depth = np.append(real_Depth, np.mean(data[edges[i]:edges[i+1]]))
cusum['Real_Start'] = real_start
cusum['Real_End'] = real_end
cusum['Real_Depth'] = real_Depth
print('Real Start =' + str(self.t[cusum['Real_Start']] ) )
print('Real End =' + str(self.t[cusum['Real_End']] ) )
cusum['CurrentLevels'] = [np.average(data[edges[i]+minlength:edges[i+1]]) for i in range(nStates)] #detect current levels during detected sub-event
print('Length of time = ' + str(len(self.t)))
print('Edges[-1] = ' + str(edges[-1]))
cusum['EventDelay'] = edges #locations of sub-events in the data
cusum['Threshold'] = threshhold #record the threshold used
print('Event = '+str( cusum['EventDelay']))
cusum['jumps'] = np.diff(cusum['CurrentLevels'])
#self.__recordevent(cusum)
e = timer()
print('cusum took = ' + str(e-s) + 's')
return cusum
def print_fitting(self, cusum):
self.p1.plot(self.t[0: cusum['EventDelay'][1]], np.repeat(np.array([cusum['CurrentLevels'][0]]), len(self.t[0:cusum['EventDelay'][1]])), pen=pg.mkPen(color=(173, 27, 183), width=3))
for number_cusum in np.arange(1,len(cusum['EventDelay'])-1 ,1):
self.p1.plot(self.t[cusum['EventDelay'][number_cusum]:cusum['EventDelay'][number_cusum+1]], np.repeat(np.array([cusum['CurrentLevels'][number_cusum]]), len(self.t[cusum['EventDelay'][number_cusum]:cusum['EventDelay'][number_cusum+1]])), pen=pg.mkPen(color=(173, 27, 183), width=3))
self.p1.plot(np.linspace(self.t[cusum['EventDelay'][number_cusum]],self.t[cusum['EventDelay'][number_cusum]] + 0.00001 ,100),np.linspace(cusum['CurrentLevels'][number_cusum-1],cusum['CurrentLevels'][number_cusum],100), pen=pg.mkPen(color=(173, 27, 183), width=3))
self.p1.plot(self.t[cusum['EventDelay'][-2]:-1], np.repeat(np.array([cusum['CurrentLevels'][-1]]), len(self.t[cusum['EventDelay'][-2]:-1])), pen=pg.mkPen(color=(173, 27, 183), width=3))
self.p1.autoRange()

Event Timeline

c4science · Help