def test_sem(self):
# This is not in R, so used: sqrt(var(testcase)*3/4) / sqrt(3)
y = mstats.sem(self.testcase)
assert_almost_equal(y, 0.6454972244)
n = self.testcase.count()
assert_allclose(mstats.sem(self.testcase, ddof=0) * np.sqrt(n/(n-2)),
mstats.sem(self.testcase, ddof=2))
def test_sem(self):
# This is not in R, so used: sqrt(var(testcase)*3/4) / sqrt(3)
y = mstats.sem(self.testcase)
assert_almost_equal(y, 0.6454972244)
n = self.testcase.count()
assert_allclose(mstats.sem(self.testcase, ddof=0) * np.sqrt(n/(n-2)),
mstats.sem(self.testcase, ddof=2))
def test_sem(self):
# This is not in R, so used: sqrt(var(testcase)*3/4) / sqrt(3)
# Note, differs from stats.sem return due to different ddof (backwards
# compat reasons).
y = mstats.sem(self.testcase)
assert_almost_equal(y, 0.55901699437494745)
n = self.testcase.count()
assert_allclose(mstats.sem(self.testcase, ddof=0) * np.sqrt(n/(n-2)),
mstats.sem(self.testcase, ddof=2))
def test_sem(self):
"""
this is not in R, so used
sqrt(var(testcase)*3/4)/sqrt(3)
"""
#y = stats.sem(self.shoes[0])
#assert_approx_equal(y,0.775177399)
y = mstats.sem(self.testcase)
assert_almost_equal(y,0.6454972244)
def test_sem(self):
"""
this is not in R, so used
sqrt(var(testcase)*3/4)/sqrt(3)
"""
#y = stats.sem(self.shoes[0])
#assert_approx_equal(y,0.775177399)
y = mstats.sem(self.testcase)
assert_almost_equal(y, 0.6454972244)
def test_sem(self):
# This is not in R, so used: sqrt(var(testcase)*3/4) / sqrt(3)
y = mstats.sem(self.testcase,ddof=1)
assert_almost_equal(y,0.6454972244)
def test_sem(self):
# This is not in R, so used: sqrt(var(testcase)*3/4) / sqrt(3)
y = mstats.sem(self.testcase)
assert_almost_equal(y,0.6454972244)
def plotIt(axAll,axInd,flies,skies,baseDirs, max_num_stops, max_time_stopped):
if 'noFilter/12trials' in baseDirs:
individualToPlot = 1#2
else:
individualToPlot = 4#2
PLOTCOLORS = {'diffuserPol/12trials':'k','diffuserPol/sham12trials':'b', 'polDiffuser/12trials':'r','noFilter/12trials':'k', 'noFilter/sham12trials':'b','diffuser/12trials':'r'}
BASEDIR_ORDER = [5,3,4]
DOWNSAMPLE_RATE = 26
UNBIASED = True
numSamplesInAverages = 130*60*12*2/DOWNSAMPLE_RATE -5#still not strictly correct
#for b_ind in range(3):
for b,baseDir in enumerate(baseDirs):
#b = BASEDIR_ORDER[b_ind]
#baseDir = baseDirs[b]
numFlies = len(flies[baseDir])
results = np.ma.masked_all((numFlies,numSamplesInAverages))
for fNum in range(numFlies):
fly = flies[baseDir][fNum].copy()
sky = skies[baseDir][fNum].copy()
orientationsAll = np.ma.masked_invalid(fly['orientations'],copy=True)
timesAll = np.ma.masked_invalid(fly['times'],copy=True)
timesAll.mask = orientationsAll.mask
orientations = orientationsAll.compressed()
times = timesAll.compressed()
if baseDir in ['diffuserPol/12trials','diffuserPol/sham12trials', 'polDiffuser/12trials']:
p=2
orientations = p*orientations
YMININD, YMAXIND = -.2,.4
NUMYTICKSIND = 4
YMINALL, YMAXALL = 0, .4
else:
YMININD, YMAXIND = .5,1
NUMYTICKSIND = 6
YMINALL, YMAXALL = 0, .5
numStops = get_num_stops(fly, sky['changeTimes'][0], sky['changeTimes'][-1])
timeStopped = get_time_stopped(fly, sky['changeTimes'][0], sky['changeTimes'][-1])
if numStops <= max_num_stops and timeStopped <= max_time_stopped:
experimentInds = (times >= sky['changeTimes'][0]) & (times <= sky['changeTimes'][-1])
experimentOrientations = orientations[experimentInds].copy()
experimentTimes = times[experimentInds].copy()
downsampledOrientations = experimentOrientations[::DOWNSAMPLE_RATE]
experimentStartTime = experimentTimes[0]
timesMinusStart = experimentTimes - experimentStartTime
downsampledTimes = timesMinusStart[::DOWNSAMPLE_RATE]
orientationsRad = downsampledOrientations*np.pi/180.0
complexOrientations = np.exp(np.complex(0,1)*orientationsRad)
meanComplexOrientation = np.mean(complexOrientations)/abs(np.mean(complexOrientations))
meanOrientation = np.arctan2(meanComplexOrientation.imag,meanComplexOrientation.real)
#complexOrientationsMinusMean = np.exp(np.complex(0,1)*(orientationsRad-meanOrientation))
#complexOrientationsMinusMean = complexOrientations / meanComplexOrientation
complexOrientationsMinusMean = complexOrientations
resultTimes = np.concatenate((-downsampledTimes[::-1],downsampledTimes[1:]))
try:
result = np.correlate(complexOrientationsMinusMean, complexOrientationsMinusMean, mode='full',old_behavior=False)
except TypeError:
print "running old numpy version"
result = np.correlate(complexOrientationsMinusMean, complexOrientationsMinusMean.conjugate(), mode='full')
result = result.real
#result = abs(result)
N = (resultTimes.size+1)/2.0
if UNBIASED:
bias = 1.0/(N - abs(np.arange(-N+1,N)))
result = result*bias
else:
result = result/N
results[fNum,:] = result[:numSamplesInAverages]
resultTimesMin = resultTimes/60.0
#lines = axAll.plot(resultTimesMin,result)
if (baseDir == 'noFilter/12trials' or baseDir == 'diffuserPol/12trials') and fNum == individualToPlot:
lines = axInd.plot(resultTimesMin,result,'k')
avgLines = axAll.plot(resultTimesMin[:numSamplesInAverages],results.mean(0),color=PLOTCOLORS[baseDir],linewidth=1)
#avgLines = flyTools.plot_mean_sem(axAll,results,resultTimes[:numSamplesInAverages])
avgLine = avgLines[0]
avgLine.set_label(baseDir)
avgLine.set_zorder(BASEDIR_ORDER[b])
#sqrtNumFliesIncluded = np.sqrt(len(results[:,0].compressed()))
#patchHandle = axAll.fill_between(resultTimesMin[:numSamplesInAverages],results.mean(0)-results.var(0),results.mean(0)+results.var(0))
patchHandle = axAll.fill_between(resultTimesMin[:numSamplesInAverages],results.mean(0)-sem(results,0),results.mean(0)+sem(results,0))
patchHandle.set_facecolor([.8, .8, .8])
patchHandle.set_edgecolor('none')
for ax in [axAll,axInd]:
for x in np.linspace(-12,12,13):
ax.axvline(x,color='g')
XMIN,XMAX = -8,8
ax.set_xlim((XMIN-.5,XMAX+.5))
ax.spines['left'].set_position(('outward',8))
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
try:
ax.spines['bottom'].set_bounds(XMIN,XMAX)
except AttributeError:
print "running old matplotlib version"
ax.xaxis.set_ticks_position('bottom')
axInd.spines['bottom'].set_visible(False)
axInd.set_xticklabels([],visible=False)
axInd.set_xticks([])
axInd.set_ylim((YMININD,YMAXIND))
axInd.yaxis.set_ticks_position('left')
axInd.set_yticks(np.linspace(YMININD,YMAXIND,NUMYTICKSIND))
axInd.set_ylabel('Autocorrelation')
axAll.spines['bottom'].set_position(('outward',8))
axAll.set_xticks(np.arange(XMIN,XMAX+1,2))
axAll.set_xlabel('Time (min)')
axAll.set_ylim((YMINALL,YMAXALL))
axAll.yaxis.set_ticks_position('left')
axAll.set_yticks(np.linspace(YMINALL,YMAXALL,3))
axAll.set_ylabel('Autocorrelation')
for ax in [axAll,axInd]:
for tickLine in ax.get_xticklines() + ax.get_yticklines():
tickLine.set_markeredgewidth(1)
def plotIt(axesList,flies,skies,baseDirs, max_num_stops, max_time_stopped):
#TITLES = {'diffuserPol/12trials':'Polarizer under \ndiffuser','diffuserPol/sham12trials':'No switching', 'polDiffuser/12trials':'Diffuser under \npolarizer','noFilter/12trials':'Switching', 'noFilter/sham12trials':'No switching','diffuser/12trials':'Diffuser'}
TITLES = {'diffuserPol/12trials':'Polarized','diffuserPol/sham12trials':'No switching', 'polDiffuser/12trials':'Unpolarized','noFilter/12trials':'Switching', 'noFilter/sham12trials':'No switching','diffuser/12trials':'Diffuser'}
SECONDS_BEFORE = 10
SECONDS_AFTER = 10
SAMPLING_RATE = 125
NUM_SAMPLES = (SECONDS_AFTER + SECONDS_BEFORE)*SAMPLING_RATE
#NUM_SAMPLES = 2000
NUM_SWITCHES = 12
MIN_EXPERIMENT_TIME = 11
#MAX_NUM_STOPS = 4
#MAX_TIME_STOPPED = 60
SMOOTH_WINDOW_LENGTH = 65
FONTDICT = {'fontsize':8}
PLOTCOLORS = ['k','b','r','g','y','c']
TIME_BEFORE_TO_AVERAGE = 10
TIME_AFTER_TO_AVERAGE = 10
allSpeedChanges = list()
#fig = pylab.figure()
ax1 = axesList[3]
for baseDirInd, baseDir in enumerate(baseDirs):
numFlies = len(flies[baseDir])
velocities = np.ma.masked_all((numFlies,NUM_SAMPLES,NUM_SWITCHES))
times = np.ma.masked_all((numFlies,NUM_SAMPLES,NUM_SWITCHES))
speedChange = np.ma.masked_all((numFlies,NUM_SWITCHES))
for flyInd in range(numFlies):
fly = flies[baseDir][flyInd].copy()
sky = skies[baseDir][flyInd].copy()
if fly['times'][-1] - fly['times'][0] > MIN_EXPERIMENT_TIME*60:
alpha = np.ma.masked_invalid(np.mod(-fly['orientations']*np.pi/180.0 + np.pi/2 + np.pi,2*np.pi),copy=True)
time = np.ma.masked_invalid(fly['times'],copy=True)
complexAlpha = np.exp(np.complex(0,1)*alpha)
smoothedAlpha = flat_smooth(complexAlpha,SMOOTH_WINDOW_LENGTH)
smoothedAlpha = smoothedAlpha/abs(smoothedAlpha) # should be unnecessary
#smoothedAngularVel = (smoothedAlpha[1:]/smoothedAlpha[:-1])/np.diff(time)
##smoothedAngularVel = (smoothedAlpha[1:]/smoothedAlpha[:-1])/np.mean(np.diff(time))
a1 = np.arctan2(smoothedAlpha[1:].imag,smoothedAlpha[1:].real)
a0 = np.arctan2(smoothedAlpha[:-1].imag,smoothedAlpha[:-1].real)
smoothedAngularVel = np.exp(np.complex(0,1)*(a1-a0))
smoothedAngularVelToPlot = np.arctan2(smoothedAngularVel.imag,smoothedAngularVel.real)/np.mean(np.diff(time))
#smoothedAngularVel = flat_smooth(complexAlpha[1:]/complexAlpha[:-1],SMOOTH_WINDOW_LENGTH)/np.mean(np.diff(time))
#smoothedAngularVel = complexAlpha[1:]/complexAlpha[:-1]/np.mean(np.diff(time))
##smoothedAngularVelToPlot = np.arctan2(smoothedAngularVel.imag,smoothedAngularVel.real)
#smoothedAngularVelToPlot = flat_smooth(np.arctan2(smoothedAngularVel.imag,smoothedAngularVel.real),SMOOTH_WINDOW_LENGTH)
experimentStartTime = sky['changeTimes'][0]
experimentEndTime = sky['changeTimes'][12]
timeStopped = get_time_stopped(fly,experimentStartTime,experimentEndTime)
numStops = get_num_stops(fly,experimentStartTime,experimentEndTime)
if timeStopped <= max_time_stopped and numStops <= max_num_stops:
for switchNumber in range(NUM_SWITCHES):
changeTime = sky['changeTimes'][switchNumber+1]
changeInd = np.argmin(abs(time - changeTime))
startTime = changeTime - SECONDS_BEFORE
startInd = np.argmin(abs(time - startTime))
endTime = changeTime + SECONDS_AFTER
endInd = np.argmin(abs(time - endTime))
indicesToSample = np.arange(startInd,endInd,dtype=int)
#indicesToSample = np.array(np.round(np.linspace(startInd,endInd,NUM_SAMPLES)),dtype=int)
numStopsDuring = get_num_stops(fly,startTime,endTime)
if numStopsDuring == 0 and fly['times'][-1]>endTime:
startAvgTime = changeTime-TIME_BEFORE_TO_AVERAGE
startAvgInd = np.argmin(abs(time - startAvgTime))
endAvgTime = changeTime+TIME_AFTER_TO_AVERAGE
endAvgInd = np.argmin(abs(time - endAvgTime))
avgSpeedAfter = np.mean(abs(smoothedAngularVelToPlot[changeInd:endAvgInd]))
#absAvgAfter = avgAfter + np.complex(0,1)*abs(avgAfter.imag) - np.complex(0,1)*avgAfter.imag
avgSpeedBefore = np.mean(abs(smoothedAngularVelToPlot[startAvgInd:changeInd]))
#absAvgBefore = avgBefore + np.complex(0,1)*abs(avgBefore.imag) - np.complex(0,1)*avgBefore.imag
velocities[flyInd,:,switchNumber] = abs(smoothedAngularVelToPlot[indicesToSample[:NUM_SAMPLES]])
times[flyInd,:,switchNumber] = time[indicesToSample[:NUM_SAMPLES]] - time[changeInd]
speedChange[flyInd,switchNumber] = avgSpeedAfter - avgSpeedBefore
PLOT_STEP = 13
speeds = abs(velocities)
#speeds = velocities*np.sign(velocities)
#flySpeeds = np.mean(speeds,2)
flySpeedsDegPerSec = np.mean(speeds,2)*180/np.pi
flyTimes = np.mean(times,2)
ax = axesList[baseDirInd]
meanSpeed = np.mean(flySpeedsDegPerSec,0)[::PLOT_STEP]
#semToPlot = np.std(flySpeedsDegPerSec,0)[::PLOT_STEP]/np.sqrt(len(flySpeedsDegPerSec[:,0].compressed()))
semToPlot = sem(flySpeedsDegPerSec,0)[::PLOT_STEP]
stdToPlot = np.std(flySpeedsDegPerSec,0)[::PLOT_STEP]
#lineHandle = ax.plot(flyTimes.transpose(),flySpeedsDegPerSec.transpose(),'k')
lineHandle = ax.plot(np.mean(flyTimes,0)[::PLOT_STEP],meanSpeed)
lineHandle[0].set_color(PLOTCOLORS[baseDirInd])
#patchHandle = ax.fill_between(np.mean(flyTimes,0)[::PLOT_STEP],meanSpeed-stdToPlot,meanSpeed+stdToPlot)
#patchHandle = ax.fill_between(np.mean(flyTimes,0)[::PLOT_STEP],meanSpeed-semToPlot,meanSpeed+semToPlot)
#patchHandle.set_facecolor([.4, .4, .4])
#patchHandle.set_edgecolor('none')
ax.set_title(TITLES[baseDir],FONTDICT)
if 'noFilter/12trials' in baseDirs:
YMIN, YMAX = 20, 70
YSPINEMIN,YSPINEMAX = 30, 50
##YMIN, YMAX = .15, .55
##YSPINEMIN,YSPINEMAX = .2, .4
else:
YMIN, YMAX = 15, 50
YSPINEMIN,YSPINEMAX = 20, 40
##YMIN, YMAX = .1, .35
##YSPINEMIN,YSPINEMAX = .2, .3
ax.set_ylim((YMIN,YMAX))
ax.set_xlim((-SECONDS_BEFORE,SECONDS_AFTER))
ax.spines['left'].set_position(('outward',8))
ax.spines['bottom'].set_position(('outward',0))
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
ax.yaxis.set_ticks_position('left')
ax.xaxis.set_ticks_position('bottom')
ax.set_xticks(np.linspace(-SECONDS_BEFORE,SECONDS_AFTER,3))
ax.set_xlabel('Time (s)')
if baseDirInd == 0:
#ax.set_ylabel('Angular speed (deg s' + r'$^{-1} $' + ')')
ax.set_ylabel('Angular speed (deg s-1)')
try:
ax.spines['left'].set_bounds(YSPINEMIN,YSPINEMAX)
ax.set_yticks(np.linspace(YSPINEMIN,YSPINEMAX,2))
except AttributeError:
print "running old matplotlib version"
#ax.set_yticks(np.linspace(0,YMAX,NUM_YTICKS))
else:
ax.set_yticklabels([],visible=False)
ax.set_yticks([])
ax.spines['left'].set_visible((False))
for tickLine in ax.get_xticklines() + ax.get_yticklines():
tickLine.set_markeredgewidth(1)
flySpeedChange = np.mean(speedChange,1)
flySpeedChangeDegPerSec = flySpeedChange*180/np.pi
boxplotHandle = ax1.boxplot(flySpeedChangeDegPerSec.compressed(),positions=[baseDirInd])
boxHandle = boxplotHandle['boxes'][0]
boxHandle.set_color(PLOTCOLORS[baseDirInd])
for whiskerHandle in boxplotHandle['whiskers']:
whiskerHandle.set_linestyle('-')
whiskerHandle.set_color('k')
for capHandle in boxplotHandle['caps']:
capHandle.set_visible(False)
for flierHandle in boxplotHandle['fliers']:
flierHandle.set_marker('+')
flierHandle.set_markerfacecolor(PLOTCOLORS[baseDirInd])
flierHandle.set_markeredgecolor(PLOTCOLORS[baseDirInd])
allSpeedChanges.append(flySpeedChangeDegPerSec.compressed())
ax1.set_xlim((-.5,len(baseDirs)-.5))
#YMIN = -.125
#YMAX = .25
#ax1.set_ylim((YMIN-.0001,YMAX))
ax1.spines['left'].set_position(('outward',2))
for baseDirInd, speedChanges in enumerate(allSpeedChanges):
t, p_val = ttest_1samp(speedChanges,0)
if p_val < .001:
ax1.text(baseDirInd-.25,ax1.get_ylim()[1],'***')
elif p_val < .01:
ax1.text(baseDirInd-.25,ax1.get_ylim()[1],'**')
elif p_val < .05:
ax1.text(baseDirInd-.25,ax1.get_ylim()[1],'*')
elif p_val >=.05:
ax1.text(baseDirInd-.25,ax1.get_ylim()[1],'NS')
ax1.spines['bottom'].set_visible(False)
ax1.spines['top'].set_visible(False)
ax1.spines['right'].set_visible(False)
try:
ax1.spines['left'].set_bounds(-10,10)
ax1.set_yticks(np.linspace(-10,10,3))
except AttributeError:
print "running old matplotlib version"
ax1.yaxis.set_ticks_position('left')
#ax1.set_yticks(np.linspace(YMIN,YMAX,4))
#ax1.set_yticklabels(ax1.get_yticks())
for tickLine in ax1.get_yticklines():
tickLine.set_markeredgewidth(1)
#ax1.set_xticks([])
if 'noFilter/12trials' in baseDirs:
XLABELS = ['S','N','D']
else:
XLABELS = ['P','N','U']
ax1.xaxis.set_ticks_position('bottom')
ax1.set_xticks(range(3)) #axAll.set_xticks([])
ax1.set_xticklabels(XLABELS)
for ax in axesList[:3]:
switchPatchHandle = ax.axvspan(0, SECONDS_AFTER)
switchPatchHandle.set_facecolor([.8,.8,.8])
switchPatchHandle.set_edgecolor('none')
switchPatchHandle.set_zorder(-1)
return allSpeedChanges
