def glmer(self, formula, family, lmerVar="glmer1"):
if os.path.exists(".rbridge-glmer.csv"):
os.remove(".rbridge-glmer.csv")
# Peform lmer
self.write("library(lmerTest)")
s = self.call("(%s <- glmer(%s, family=%s))" % (lmerVar, formula, family))
self.write('write.csv(summary(%s)$coef, ".rbridge-glmer.csv")' % lmerVar)
while not os.path.exists(".rbridge-glmer.csv"):
time.sleep(0.1)
# Try this a few times, because sometimes the csv hasn't been written
# yet
for i in range(10):
try:
dm = CsvReader(".rbridge-glmer.csv").dataMatrix()
break
except:
time.sleep(1)
dm.rename("f0", "effect")
dm.rename("Estimate", "est")
dm.rename("Std. Error", "se")
dm.rename("Z value", "z")
dm.rename("Pr(>|z|)", "p")
dm = dm.addField("estProb", dtype=float)
dm["estProb"] = 1.0 / (1.0 + np.exp(-dm["est"]))
return dm
def refixLaunch(exp,corr=False, trim=True, bins=8):
"""
"""
src = 'selected_dm_%s_WITH_drift_corr_onlyControl_True.csv' % exp
dm = CsvReader(src).dataMatrix()
lat1 = "saccLat1"
lat2 = "saccLat2"
if corr:
fix1 = "endX1CorrNormToHandle"
fix2 = "endX2CorrNormToHandle"
else:
fix1 = "endX1NormToHandle"
fix2 = "endX2NormToHandle"
for var in [lat1, lat2, fix1, fix2]:
dm = dm.select("%s != ''" % var)
# Trim the data
if trim:
dm = dm.selectByStdDev(["file"],lat2,
verbose=False)
dm = dm.removeField("__dummyCond__")
dm = dm.removeField("__stdOutlier__")
dm = dm.selectByStdDev(["file"], fix2,
verbose=False)
dm = dm.addField('saccLat2_perc', dtype=float)
dm = dm.calcPerc(lat2, 'saccLat2_perc', nBin=bins)
cmLat1 = dm.collapse(['saccLat2_perc'], lat1)
cmX = cmLat2 = dm.collapse(['saccLat2_perc'], lat2)
cmFix1 = dm.collapse(['saccLat2_perc'], fix1)
cmFix2 = dm.collapse(['saccLat2_perc'], fix2)
fig = plt.figure(figsize = (3,6))
plt.subplots_adjust(left=.2, bottom=.15)
colList = [orange[1], blue[1]]
nRows = 3
nCols = 1
nPlot = 0
lTitles = ["Landing pos 2", "Landing pos 1"]
lTitles.reverse()
for cmY in [cmFix2, cmFix1]:#, cmLat1]:
#nPlot +=1
#plt.subplot(nRows, nCols, nPlot)
color = colList.pop()
plt.plot(cmX['mean'], cmY['mean'], marker = 'o', color=color, \
markerfacecolor='white', markeredgecolor=color, \
markeredgewidth=1)
plt.xlabel("Sacc lat 2")
plt.legend(lTitles, frameon=False, loc='best')
plt.axhline(0, linestyle = "--", color = gray[3])
plt.savefig("Launch_site_refixations_%s_corr_%s.png" % (exp, corr))
printSum = False
if __name__ == "__main__":
# Get dictionary containing w and h of imaginary boxes around the stimuli:
boxDict = drawBox.boxDict()
# OpenSesame output file is opened as dm. The fixation values will
# be added to this dm.
dm = CsvReader(fName, delimiter = ",").dataMatrix()
# The varialbe 'corrDirection' is determined in ascParser,
# and doesn't exist for the simulation data.
# Therefore we add it manually:
dm = dm.addField("corrDirection", dtype = str)
# Make new column headers:
for sacc in range(1,4):
# Landing position:
dm = dm.addField('endX%s' % sacc, dtype=str)
dm = dm.addField('endY%s' % sacc, dtype = str)
# Normalised landing position:
dm = dm.addField('endX%sNorm' % sacc, dtype = str)
dm = dm.addField('endY%sNorm' % sacc, dtype = str)
# Landing position where positive indicates ToHandle"
dm = dm.addField('endX%sNormToHandle' % sacc, dtype = str)
yc = 768/2
ythr = 100
if __name__ == "__main__":
dm = CsvReader('data.csv').dataMatrix()
dm = dm.select('mask_side == "control"')
dm = dm.select('symm == "asymm"')
#dm = dm.select('gap == "overlap"')
#dm = dm.select('handle_side == "right"')
#dm = dm.select('object == "screwdriver"')
# X coordinates, where positive values indicate a deviation towards the handle
dm = dm.addField('refix1', dtype=float)
dm = dm.addField('refix2', dtype=float)
dm = dm.addField('refix3', dtype=float)
# Timestamps for each fixation
dm = dm.addField('tfix1', dtype=float)
dm = dm.addField('tfix2', dtype=float)
dm = dm.addField('tfix3', dtype=float)
for i in dm.range():
trial = dm[i]
trialNr = dm['count_trial_sequence'][i]
handleSide = trial['handle_side'][0]
print 'Trial %d (%s)' % (trialNr, handleSide)
sim = CsvReader('simulation/csv/%.4d.csv' % trialNr).dataMatrix()
def plotContrast(exp = "004B", trim=True, inclSim=True,stats=True,norm=False):
"""
Plots landing positions as a function of contrats manipulation per saccade,
for Exp 1 and Exp 2, relative to CoG.
Keyword arguments:
trim --- (default=True)
"""
# Get dm:
src = 'selected_dm_%s_WITH_drift_corr_onlyControl_False.csv' % exp
dm = CsvReader(src).dataMatrix()
colList = ["#f57900", "#3465a4"]
lLegend = ["Saccade 1", "Saccade 2"]
fig = plt.figure(figsize = (3,4))
plt.subplots_adjust(left=.2, bottom=.15)
yLim = [-.07, .12]
for sacc in ["1", "2"]:
dv = "endX%sNorm" % sacc
saccVar = "saccLat%s" % sacc
# Get dm:
# Only on-object:
dm = onObject.onObject(dm, sacc,verbose=False)
if trim:
dm = dm.removeField("__dummyCond__")
dm = dm.removeField("__stdOutlier__")
dm = dm.selectByStdDev(keys = ["file"], \
dv = dv,verbose=False)
dm = dm.removeField("__dummyCond__")
dm = dm.removeField("__stdOutlier__")
dm = dm.selectByStdDev(keys = ["file"], dv = saccVar,\
verbose=False)
# Collect mean and error bar per saccade:
lM = []
lErr = []
if norm:
# Normalize across handle side:
dm = dm.removeField("normDV")
dm= dm.addField("normDV", dtype = float)
dm= dm.withinize(dv, "normDV", \
["handle_side"], whiten=False)
dv = "normDV"
for contrast in dm.unique("contrast_side"):
contrast_dm = dm.select("contrast_side == '%s'" % contrast,\
verbose = False)
cm = contrast_dm.collapse(["file"], dv)
M = cm["mean"].mean()
SE = cm['mean'].std() / np.sqrt(len(cm))
CI = SE * critVal
lM.append(M)
lErr.append(CI)
if stats:
# Run a full LME
print "Exp = ", exp
print "DV = ", dv
print "trim = ", trim
print "norm = ", norm
print "saccVar = ", saccVar
lmeContrast(R, dm, dv, saccVar, exp=exp)
#raw_input()
xData = range(len(lM))
yData = lM
yErr = lErr
col = colList.pop()
plt.errorbar(xData, yData, yerr=yErr, fmt='o-', marker = "o", \
color = col, markerfacecolor='white', markeredgecolor=col, \
markeredgewidth=1)
plt.axhline(0, linestyle = "--")
plt.ylim(yLim)
plt.ylabel(yTitle)
#ax.yaxis.set_ticklabels([])
plt.legend(lLegend, frameon = False)
plt.axhline(0, color = "black", linestyle = "--")
plt.xlabel("High-contrast side")
spacing = 0.5
xTicks = range(0,3)
xLabels = ["Left", "Control", "Right"]
plt.xticks(xTicks, xLabels, rotation = .5)
plt.xlim(min(xTicks)-spacing, max(xTicks)+spacing)
plt.savefig(os.path.join(dst,"Contrast_Effect_%s_trim_%s_norm_%s.png") \
% (exp, trim, norm))
plt.show()
from exparser.CsvReader import CsvReader
from matplotlib import pyplot as plt
f = "Sylvie.csv"
dm = CsvReader(f).dataMatrix()
dm = dm.addField("trial_dur")
for i in dm.range():
# Skip the very first trial because it doesn't have a previous trial:
if i == 0:
dm["trial_dur"][i] = -1000
else:
s = dm["time_correct_response"][i-1]
e = dm["time_correct_response"][i]
dur = e-s
dm["trial_dur"][i] = dur
# Exclude first trial, which was skipped:
dm = dm.select("trial_dur > 0")
dm["trial_dur"] = dm["trial_dur"]/1000/60
#import analyse
#analyse.plotDist(dm, "trial_dur", bins = 100)
#plt.show()
plt.hist(dm["trial_dur"], bins=100)
plt.show()
# Exclude trials that took longer than 1 minute, because this was probably a
def ovp(exp,corr=False, trim=True, bins=8):
"""
"""
src = 'selected_dm_%s_WITH_drift_corr_onlyControl_True.csv' % exp
dm = CsvReader(src).dataMatrix()
lat1 = "saccLat1"
lat2 = "saccLat2"
prob = "saccCount"
dur1 = "durationFix1"
dur2 = "durationFix2"
total = "gazeDur"
rt = "rtFromStim"
if corr:
fix1 = "endX1CorrNormToHandle"
fix2 = "endX2CorrNormToHandle"
else:
fix1 = "endX1NormToHandle"
fix2 = "endX2NormToHandle"
for var in [lat1, lat2, fix1, fix2, dur1, dur2, total, rt]:
dm = dm.select("%s != ''" % var)
# Trim the data
if trim:
dm = dm.selectByStdDev(["file"],lat2,
verbose=False)
dm = dm.removeField("__dummyCond__")
dm = dm.removeField("__stdOutlier__")
dm = dm.selectByStdDev(["file"], fix2,
verbose=False)
dm = dm.addField('land2_perc', dtype=float)
dm = dm.calcPerc(fix2, 'land2_perc', nBin=bins)
cmLat1 = dm.collapse(['land2_perc'], lat1)
cmLat2 = dm.collapse(['land2_perc'], lat2)
cmFix1 = dm.collapse(['land2_perc'], fix1)
cmX = cmFix2 = dm.collapse(['land2_perc'], fix2)
cmProb = dm.collapse(['land2_perc'], prob)
cmDur1 = dm.collapse(['land2_perc'], dur1)
cmDur2 = dm.collapse(['land2_perc'], dur2)
cmTotal = dm.collapse(['land2_perc'], total)
cmRt = dm.collapse(['land2_perc'], rt)
fig = plt.figure(figsize = (3,10))
plt.subplots_adjust(left=.2, bottom=.15, hspace = .2)
lTitles = ["dur 1", "prob", "land 1", "rt", "total", "lat 1", "lat 2", "dur 2"]
lTitles.reverse()
nRows = len(lTitles)
nCols = 1
nPlot = 0
for cmY in [cmDur1, cmProb, cmFix1, cmRt, cmTotal, \
cmLat1, cmLat2, cmDur2]:
nPlot +=1
ax = plt.subplot(nRows, nCols, nPlot)
#color = colList.pop()
color = blue[1]
plt.plot(cmX['mean'], cmY['mean'], marker = 'o', color=color, \
markerfacecolor='white', markeredgecolor=color, \
markeredgewidth=1)
plt.ylabel(lTitles.pop())
plt.axvline(0, linestyle = "--", color = gray[3])
if nPlot == nRows:
plt.xlabel("Second landing position")
else:
ax.xaxis.set_ticklabels([])
plt.savefig("ovp2_%s_corr_%s.png" % (exp, corr))
def plotGap(exp, dv = "saccLat1", trim = False, norm=True, nBins = 50, \
exclFastSacc = False):
"""
"""
src = 'selected_dm_%s_WITH_drift_corr_onlyControl_True.csv' % exp
dm = CsvReader(src).dataMatrix()
# Determine sacc:
sacc = [int(x.group()) for x in re.finditer(r'\d+', dv)][0]
dm = onObject.onObject(dm, sacc)
# Exclude very fast saccades:
dm = dm.select("%s > 80" % dv)
colList = [orange[1], blue[1]]
# Trim the data
if trim:
dm = dm.removeField("__dummyCond__")
dm = dm.removeField("__stdOutlier__")
dm = dm.selectByStdDev(["file"], dv, \
verbose=False)
dm = dm.removeField("__dummyCond__")
dm = dm.removeField("__stdOutlier__")
dm = dm.selectByStdDev(["file"], "endX%sNormToHandle" % sacc, \
verbose=False)
if exp == "004A":
dm = dm.removeField("__dummyCond__")
dm = dm.removeField("__stdOutlier__")
dm = dm.selectByStdDev(["file"], "endX%sCorrNormToHandle" % sacc, \
verbose=False)
else:
dm = dm.select("%s < 1000" % dv)
# Normalize saccade latencies
if norm:
dm= dm.addField("normSacc", dtype = float)
dm= dm.withinize(dv, "normSacc", \
["file"], whiten=False)
dv = "normSacc"
for gap in dm.unique("gap"):
_dm = dm.select("gap == '%s'" % gap)
samp = _dm[dv]
y, edges = np.histogram(samp, bins = nBins)
y = normY(y)
x = .5*edges[1:] + .5*edges[:-1]
col = colList.pop()
plt.plot(x, y, color = col)
plt.fill_between(x, 0, y, alpha=.3, color=col)
plt.legend(dm.unique("gap"))
plt.ylim([0,1.1])
def dmSim():
"""
"""
dm = CsvReader('./simulation/data.csv').dataMatrix()
# X coordinates, where positive values indicate a deviation towards the handle
dm = dm.addField('sacc1_ex', dtype=float)
dm = dm.addField('sacc2_ex', dtype=float)
dm = dm.addField('sacc3_ex', dtype=float)
dm = dm.addField('sacc1_ey', dtype=float)
dm = dm.addField('sacc2_ey', dtype=float)
dm = dm.addField('sacc3_ey', dtype=float)
dm = dm.addField("trialId", dtype = str)
# Timestamps for each fixation
dm = dm.addField('saccLat1', dtype=float)
dm = dm.addField('saccLat2', dtype=float)
dm = dm.addField('saccLat3', dtype=float)
for i in dm.range():
trial = dm[i]
trialNr = dm['count_trial_sequence'][i]
dm["trialId"][i] = trialNr
handleSide = trial['flip'][0]
print 'Trial %d (%s)' % (trialNr, handleSide)
sim = CsvReader('simulation/simulation/csv/%.4d.csv' % trialNr).dataMatrix()
fixNr = 1
for fix in sim:
t = fix['t'][0]
dx = fix['x'][0]
dy = fix['y'][0]
dm["sacc%s_ey" % fixNr][i] = dy
dm['sacc%d_ex' % fixNr][i] = dx
dm['saccLat%d' % fixNr][i] = t
fixNr += 1
if fixNr > 3:
break
dm.save('dm_simulation.csv')
dm = dm.addField("expId", dtype = str, default = "sim")
dm = dm.addField("symm", dtype = str, default = "asymm")
dm = dm.addField("saccCount", default = 3)
dm = dm.addField("mask_side", dtype = str, default = "control")
dm = dm.addField("cond", dtype = str, default = "not_practice")
dm = dm.addField("rep", dtype = str, default = "not_practice")
dm = dm.addField("checkFixDotFailed", dtype = str, default = "False")
dm = dm.addField("checkObjectFailed", dtype = str, default = "False")
dm = dm.addField("file", dtype = str, default = "sim")
return dm
