def simon(dm):
"""
"""
pm = PivotMatrix(dm, ['response_hand', 'mask_side'], ['file'], dv='RT', colsWithin=False)
pm._print()
pm.barPlot(show=True)
plt.title("simon effect")
def affordance(dm):
"""
"""
dm = dm.select('symm == "asymm"')
pm = PivotMatrix(dm, ['response_hand', 'handle_side'],['file'], dv = "RT")
pm.barPlot(show=True)
plt.title("affordance effect")
def gapEffect(dm, dv = "saccLat"):
"""
Regardless of color change...
"""
pm = PivotMatrix(dm, ['gap'],['file'],dv =dv)
pm._print()
pm.barPlot(show=True)
plt.ylabel(dv)
plt.title("gap effect")
def affLumCongr(dm):
dm = dm.select('symm == "asymm"')
xLabels = ["hande left", "handle right"]
lLabels = ["control", "more contrast right", "more contrast left"]
yLabel = "landing position: negative = left, positive = right"
pm = PivotMatrix(dm, ['handle_side', 'mask_side'], ['file'], \
dv='endX')
pm._print()
pm.barPlot(xLabels = xLabels, lLabels = lLabels, yLabel = yLabel, show=True)
plt.title("overlap between contrast and affordance")
def vfDiff(dm,dv = "RT"):
pm = PivotMatrix(dm, ['visual_field'],['file'], dv = dv)
pm.barPlot(show=True)
plt.ylabel(dv)
plt.title("visual field difference")
def ovp(dm, rt = "rtFromLanding", ws = True, nRows = 1, nCols = 8, nBin = 5, \
binVar = "endX1Norm", binVar2 = None, figSize = (15,4), onlyControl = True):
"""
Plots OVP effects as a function of binned initial landing position.
Arguments:
dm --- data matrix
rt --- variable for RT
ws --- Vincentize or not.
binVar --- variable for first row of OVP plots
binVar2 --- variable for second row of OVP plots (optional)
onlyControl --- exclude contrast manipulation. Default = True
"""
# If wanted, exclude the contrast manipulation:
if onlyControl:
dm = dm.select("mask_side == 'control'")
# Determine whether or not to first split data per participant
# (i.e. to make within-subjects bins):
if ws:
keys = ['file']
else:
keys = None
# Determine exp:
exp = dm["exp"][0]
# Add factors containing bins:
binnedVar = "binned_%s"%binVar
dm = dm.addField(binnedVar)
dm = dm.calcPerc(binVar, binnedVar,keys = keys, nBin = nBin)
factorList = [binnedVar]
if binVar2 != None:
binnedVar2 = "binned_%s" % binVar2
dm = dm.addField(binnedVar2)
dm = dm.calcPerc(binVar2, binnedVar2, keys = keys, nBin = nBin)
factorList.append(binnedVar2)
# Create figure:
fig = plt.figure(figsize = figSize)
plt.subplots_adjust(wspace = .5)
figName = "%s: OVP effects as a function of binned %s - WS = %s" % (exp, factorList, ws)
plt.suptitle(figName)
plotCount = 0
# Define keys:
subjectID = ["file"]
dvList = [rt, "durationFix1", "refixProb", "saccCount", "gazeDur", "durationFix2", \
"endX2Norm", "saccLat1"]
for factors in factorList:
for dv in dvList:
plotCount +=1
# Exclude trials on which the dv does not exist:
_dm = dm.select("%s != ''"%dv)
plt.subplot(nRows, nCols, plotCount)
plt.title(dv)
pm = PivotMatrix(_dm, factors, subjectID, dv, colsWithin = True)
pm.barPlot(fig = fig)
if dv == 'endX2Norm':
plt.axhline(0, linestyle = '--', color = 'gray')
# Determine ylim's:
if exp == "004A":
if dv == "rtFromLanding":
plt.ylim(500,550)
if dv == "durationFix1":
plt.ylim(260, 300)
if dv == "durationFix2":
plt.ylim(315, 355)
if dv == "refixProb":
plt.ylim(0.58, 0.72)
if dv == "saccCount":
plt.ylim(1.75, 1.94)
if dv == "endX2Norm":
plt.ylim(-.2, .2)
if dv == "gazeDur":
plt.ylim(500, 600)
if dv == "saccLat1":
plt.ylim(150, 200)
if exp == "004B":
if dv == "rtFromLanding":
plt.ylim(400,500)
if dv == "durationFix1":
plt.ylim(260, 300)
if dv == "durationFix2":
plt.ylim(270, 300)
if dv == "refixProb":
plt.ylim(0.4, 0.65)
if dv == "saccCount":
plt.ylim(1.6, 2.)
if dv == "endX2Norm":
plt.ylim(-.2, .2)
if dv == "gazeDur":
plt.ylim(450, 510)
if dv == "saccLat1":
plt.ylim(150, 250)
#plt.ylabel(dv)
plt.xticks([])
plt.savefig("%s RT = %s nBin = %s.png"%(figName,rt, nBin))
def landWithoutContrast(factor = ["handle_side"], nRows = 3, nCols = 4, saccNr = "1", \
yLim = [-.1, .1], pp = ["file"], stats = True, colsWithin = True, sepPlots = True, exclY = True):
"""
Plots effect of handle side on landing-position variable. Only trials without
contrast manipulation are included.
Arguments:
dm
Keyword arguments:
saccNr --- nth saccade of interest (default = first, "1")
sepPlots --- save and return as separate plots or as subplots of one main figure
"""
# Create figure:
# Experiment 004A:
exp = "004A"
dm1 = getDM.getDM(exp = exp, driftCorr = True)
# Only use objects without contrast manipulation
dm1 = dm1.select("mask_side == 'control'", verbose = False)
figList = []
if not sepPlots:
fig = plt.figure(figsize = (10,2))
nRows = 1
nCols = 3
plotNr = 0
for dv in ["endX%sNorm" % saccNr, "endX%sCorrNorm" % saccNr]:
if sepPlots:
fig = plt.figure(figsize = (10,3))
figName = "exp %s dv %s colsWithin = %s - exclY = %s" % (exp, dv, colsWithin, exclY)
else:
plotNr +=1
plt.subplot(nRows,nCols,plotNr)
_dm = dm1
# Only trials where variable has a value:
_dm = _dm.select("%s != ''"%dv)
# Only on-object EM's:
_dm = _dm.select('endX%sNorm < .5' % saccNr)
_dm = _dm.select('endX%sNorm > -.5' % saccNr)
if exclY:
_dm = _dm.select('endY%sNorm < .5' % saccNr)
_dm = _dm.select('endY%sNorm > -.5' % saccNr)
# Trim:
_dm = _dm.selectByStdDev(keys = factor + pp, dv = dv)
#plt.subplot2grid((nRows,nCols), (plotCount, 0))
pm = PivotMatrix(_dm, factor, pp, dv, colsWithin=colsWithin, err='se')
#plt.axhline(0, linestyle = "--", color = "#555753")
#plt.ylim(yLim)
#pm.barPlot(fig=fig)
pm.barPlot(fig = fig, _dir='right')
plt.axvline(0, linestyle = "--", color = "#555753")
plt.axhline(0.5, linestyle = "-", color = "#555753")
plt.xlim(yLim)
if sepPlots:
if "Corr" in dv:
plt.xlabel("corrected horizontal landing position")
else:
plt.xlabel("horizontal landing position")
else:
plt.xticks([])
if stats:
am = AnovaMatrix(_dm, factors = factor, dv = dv, \
subject = pp[0])._print(maxLen=7, ret=True)
print am
#plt.subplot2grid((nRows,nCols), (plotCount, 1), colspan=3)
#plt.title("Exp = %s DV = %s" % (exp, dv))
#plt.text(0.05,0.1,am, family='monospace')
#plt.xticks([])
#plt.yticks([])
#plotCount +=1
if sepPlots:
plt.savefig("%s.svg"%figName)
plt.savefig("%s.png"%figName)
#plt.show()
figList.append(fig)
# For experiment 004B we plot only the normalised landing position.
# Correcting for CoG doesn't make sense here!
exp = "004B"
# Get dm:
dm2 = getDM.getDM(exp = exp, driftCorr = True)
# Only trials without contrast manipulation:
dm2 = dm2.select("mask_side == 'control'")
# Only trials where variable has a value:
dm2 = dm2.select("%s != ''"%dv)
# Only on-object eye movements:
dm2 = dm2.select('endX%sNorm < .5' % saccNr)
dm2= dm2.select('endX%sNorm > -.5' % saccNr)
if exclY:
dm2 = dm2.select('endY%sNorm < .5' % saccNr)
dm2= dm2.select('endY%sNorm > -.5' % saccNr)
# Pivot matrix and stats:
dv = "endX%sNorm" % saccNr
# Trim:
dm2 = dm2.selectByStdDev(keys = factor + pp, dv = dv)
if sepPlots:
fig = plt.figure(figsize = (10,3))
figName = "exp %s dv %s colsWithin = %s - exclY = %s" % (exp, dv, colsWithin, exclY)
else:
plt.subplot(nRows,nCols,3)
pm = PivotMatrix(dm2, factor, pp, dv = dv, colsWithin = colsWithin, err = 'se')
#plt.subplot2grid((nRows,nCols), (plotCount, 0))
pm.barPlot(fig = fig, _dir='right')
plt.axvline(0, linestyle = "--", color = "#555753")
plt.axhline(0.5, linestyle = "-", color = "#555753")
plt.xlim(yLim)
if sepPlots:
plt.xlabel("horizontal landing position")
else:
plt.xticks([])
if stats:
am = AnovaMatrix(dm2, factors = factor, dv = dv, \
subject = pp[0])._print(maxLen=7, ret=True)
print am
#plt.subplot2grid((nRows,nCols), (plotCount, 1), colspan=3)
#plt.title("Exp = %s DV = %s" % (exp, dv))
#plt.text(0.05,0.1,am, family='monospace')
#plt.xticks([])
#plt.yticks([])
if sepPlots:
plt.savefig("%s.png"%figName)
plt.savefig("%s.svg"%figName)
figList.append(fig)
return figList
else:
return fig
def splitHeavy(factor = ["corrDirection", "handle_side"], nRows = 3, nCols = 4, saccNr = "1", \
yLim = [-.5, .5], pp = ["file"], stats = True, colsWithin = True, sepPlots = True, exclY = True):
# Create figure:
# Experiment 004A:
exp = "004A"
dm1 = getDM.getDM(exp = exp, driftCorr = True)
dm1 = dm1.select("corrDirection != 'correction smaller than 8'")
for corr in dm1.unique("corrDirection"):
print corr
tmp = dm1.select("corrDirection == '%s'" % corr)
print "Corr direction = ", corr
print tmp.unique("object")
# Only use objects without contrast manipulation
dm1 = dm1.select("mask_side == 'control'", verbose = False)
figList = []
if not sepPlots:
fig = plt.figure(figsize = (10,2))
nRows = 1
nCols = 3
plotNr = 0
for dv in ["endX%sNorm" % saccNr, "endX%sCorrNorm" % saccNr]:
if sepPlots:
fig = plt.figure(figsize = (10,3))
figName = "split by heavy side exp %s dv %s colsWithin = %s - exclY = %s" % (exp, dv, colsWithin, exclY)
else:
plotNr +=1
plt.subplot(nRows,nCols,plotNr)
_dm = dm1
# Only trials where variable has a value:
_dm = _dm.select("%s != ''"%dv)
# Only on-object EM's
_dm = _dm.select('endX%sNorm < .5' % saccNr)
_dm = _dm.select('endX%sNorm > -.5' % saccNr)
if exclY:
_dm = _dm.select('endY%sNorm < .5' % saccNr)
_dm = _dm.select('endY%sNorm > -.5' % saccNr)
# Trim:
_dm = _dm.selectByStdDev(keys = factor + pp, dv = dv)
#plt.subplot2grid((nRows,nCols), (plotCount, 0))
pm = PivotMatrix(_dm, factor, pp, dv, colsWithin=colsWithin, err='se')
#plt.axhline(0, linestyle = "--", color = "#555753")
#plt.ylim(yLim)
#pm.barPlot(fig=fig)
pm.barPlot(fig = fig, _dir='right', legendTitle = "handle side")
plt.axvline(0, linestyle = "--", color = "#555753")
plt.axhline(0.5, linestyle = "-", color = "#555753")
plt.xlim(yLim)
if sepPlots:
if "Corr" in dv:
plt.xlabel("corrected horizontal landing position")
else:
plt.xlabel("horizontal landing position")
else:
plt.xticks([])
if stats:
am = AnovaMatrix(_dm, factors = factor, dv = dv, \
subject = pp[0])._print(maxLen=7, ret=True)
print am
#plt.subplot2grid((nRows,nCols), (plotCount, 1), colspan=3)
#plt.title("Exp = %s DV = %s" % (exp, dv))
#plt.text(0.05,0.1,am, family='monospace')
#plt.xticks([])
#plt.yticks([])
#plotCount +=1
if sepPlots:
plt.savefig("%s.svg"%figName)
plt.savefig("%s.png"%figName)
#plt.show()
figList.append(fig)
# For experiment 004B we plot only the normalised landing position.
# Correcting for CoG doesn't make sense here!
exp = "004B"
# Get dm:
dm2 = getDM.getDM(exp = exp, driftCorr = True)
# Only trials without contrast manipulation:
dm2 = dm2.select("mask_side == 'control'")
dm2 = dm2.select("corrDirection != 'correction smaller than 8'")
# Only trials where variable has a value:
dm2 = dm2.select("%s != ''"%dv)
# Only on-object eye movements:
dm2 = dm2.select('endX%sNorm < .5' % saccNr)
dm2= dm2.select('endX%sNorm > -.5' % saccNr)
if exclY:
dm2 = dm2.select('endY%sNorm < .5' % saccNr)
dm2= dm2.select('endY%sNorm > -.5' % saccNr)
# Pivot matrix and stats:
dv = "endX%sNorm" % saccNr
# Trim:
dm2 = dm2.selectByStdDev(keys = factor + pp, dv = dv)
if sepPlots:
fig = plt.figure(figsize = (10,3))
figName = "split by heavy side exp %s dv %s colsWithin = %s" % (exp, dv, colsWithin)
else:
plt.subplot(nRows,nCols,3)
pm = PivotMatrix(dm2, factor, pp, dv = dv, colsWithin = colsWithin, err = 'se')
#plt.subplot2grid((nRows,nCols), (plotCount, 0))
pm.barPlot(fig = fig, _dir='right')
plt.axvline(0, linestyle = "--", color = "#555753")
plt.axhline(0.5, linestyle = "-", color = "#555753")
plt.xlim(yLim)
if sepPlots:
plt.xlabel("horizontal landing position")
else:
plt.xticks([])
if stats:
am = AnovaMatrix(dm2, factors = factor, dv = dv, \
subject = pp[0])._print(maxLen=7, ret=True)
print am
#plt.subplot2grid((nRows,nCols), (plotCount, 1), colspan=3)
#plt.title("Exp = %s DV = %s" % (exp, dv))
#plt.text(0.05,0.1,am, family='monospace')
#plt.xticks([])
#plt.yticks([])
if sepPlots:
plt.savefig("%s.png"%figName)
plt.savefig("%s.svg"%figName)
figList.append(fig)
return figList
else:
plt.savefig('per corrDir.png')
return fig
def allSaccLats(dm, direction, figTitle = True, trim = True, cousineau = True, nBin = 15, usePm = True, err = 'se'):
"""
Throw all saccades latencies (since stim onset) in for bin analysis, irrespective of saccade count
Arguments:
dm
direction --- {"ToHandle", "ToContrast"}, variable on the y axis
Keyword arguments:
Returns fig list
"""
figList = []
print '\n\tdirection = %s \n' % direction
if direction == "ToHandle":
start_dm = dm.select("contrast_side == 'control'")
if direction == "ToContrast":
start_dm = dm.select("contrast_side != 'control'")
exp = start_dm['exp'][0]
lDm = []
if exp == "004A":
# Max sacc count = 4
lSacc = range(1,5)
if exp == "004B":
# Max sacc count = 3
lSacc = range(1,4)
for saccCount in lSacc:
# Create dms containing all latencies (i.e., the to-be-binned variable)
# of a given saccade count, e.g.
# almost all trials for the first saccade, and less and less for the
# subsequent saccades.
# Do the same for the landing positions (dv's on the y-axis):
# Select one saccade:
dm_sacc = dm.select("saccLat%s != ''" % str(saccCount))
# Create new column containing sacc count:
dm_sacc = dm_sacc.addField("saccNr", dtype = str)
dm_sacc["saccNr"] = "sacc%s" % saccCount
# Create new column containing IV and DV's, but without the
# sacc count in the column header:
dm_sacc = dm_sacc.addField("absSaccLat", dtype = float)
dm_sacc["absSaccLat"] = dm_sacc["saccLat%s" % str(saccCount)]
dm_sacc = dm_sacc.addField("abs%s" % direction, dtype = float)
dm_sacc["abs%s" % direction] = dm_sacc["endX%sNorm%s"%(str(saccCount), direction)]
dm_sacc = dm_sacc.addField("absCorr%s" % direction, dtype = float)
dm_sacc["absCorr%s" % direction] = dm_sacc["endX%sCorrNorm%s"%(str(saccCount), direction)]
# Add the dm to the list of dm's
lDm.append(dm_sacc)
# Combine all dm's into one big dm:
merged_dm = lDm[0]
for dm in lDm[1:]:
merged_dm = merged_dm + dm
corrList = ["corrected", "uncorrected"]
for corr in corrList:
fig = plt.figure(figsize = (3,7))
title = "Landing (%s) %s as a function of ALL binned sacc lats exp %s cousineau = %s trim = %s usePm = %s" \
% (corr, direction, exp, cousineau, trim, usePm)
if figTitle:
plt.title(title)
if corr == "uncorrected":
dv = "abs%s" % direction
elif corr == "corrected":
dv = "absCorr%s" % direction
print "\tDV = %s\n" % dv
# There should be no ''s in the dm anymore, but off-object saccades are still
# possible, so this filtering remains necessary:
dv_dm = onObject(merged_dm,dv)
saccLat = "absSaccLat"
# Trim the data matrix such that the most extreme latencies are excluded:
print "\n\ttrim = %s\n" % trim
if trim:
trimmed_dm = dv_dm.selectByStdDev(keys = ["file"], dv = saccLat)
else:
trimmed_dm = dv_dm
# Withinize sacc latencies, if wanted:
print "\n\tcousineau = %s\n" % cousineau
if cousineau:
_dm = trimmed_dm.addField("cousineau_%s"%saccLat, dtype = None)
_dm = _dm.withinize(saccLat, "cousineau_%s"%saccLat, "file", verbose = True, whiten=False)
saccLat = "cousineau_%s"%saccLat
else:
_dm = dv_dm
# Make bins, only for the first dv (for the second dv, the binned variable is the same and
# therefore already exists:
varToBin = saccLat
binnedVar = "binnend%s" % varToBin
binned_dm = _dm.addField(binnedVar)
binned_dm = binned_dm.calcPerc(varToBin, binnedVar ,keys = ["file"], nBin = nBin)
if not usePm:
lX = []
lY = []
for _bin in binned_dm.unique(binnedVar):
# Filter out all but one bin
dm_one_bin = binned_dm.select('%s == %f' % (binnedVar, _bin))
# Get the mean sacc lat and the mean landing position (for x and y axis, respectively):
# NOTE: withinising doesn't make any real difference for the overall pattern.
yMean = dm_one_bin[dv].mean()
xMean = dm_one_bin[varToBin].mean()
lX.append(xMean)
lY.append(yMean)
plt.plot(lX, lY, color = "#3465a4", marker = 'o')
plt.xlabel("binned saccade latencies from stimulus onset")
plt.ylabel("landing position (%s) %s" % (corr,direction))
plt.ylim(-.2, .2)
plt.axhline(0, color = "#555753", linestyle = "--", linewidth = 2)
plt.savefig(os.path.join(dst,'%s.png' % title))
if usePm:
pm = PivotMatrix(binned_dm, binnedVar, "file", dv, colsWithin = True, err = 'se')
#pm.plot(nLvl1=1, fig = fig, xLabel = "binned saccade latencies from stimulus onset", \
# yLabel = "landing position (%s) %s" % (corr,direction))
pm.barPlot(fig = fig, xLabel = "binned saccade latencies from stimulus onset", \
yLabel = "landing position (%s) %s" % (corr,direction))
plt.ylim(-.2, .2)
plt.axhline(0, color = "#555753", linestyle = "--", linewidth = 2)
plt.savefig(os.path.join(dst,'%s.png' % title))
plt.savefig(os.path.join(dst,'%s.png' % title))
figList.append(fig)
return figList
def oneFactor(dm, dv, factor, subjectID = ["file"], trim = True,
printStats = True, showFig=False, saveFig = True, figName = None,
saveExt = ".jpg", xLabel = None, xLabels = None, yLabel = None,
yLim = None, exControlMask = False, exFiller = False, exMask = False):
"""
Performs and plots a one-factorial ANOVA.
NOTE: The resulting plots are bar plots because 'pm.linePlot()' doesn't
work with one factor yet.
Arguments:
dm --- Data matrix.
dv --- Dependent variable.
factor --- Independent variable.
Keyword arguments:
subjectID --- Default = ["file"]
showFig --- Boolean indicating whether or not to show the plots. Default
= False.
saveFig --- Boolean indicating whehter or not to save the plot. Default =
True.
printStats --- TODO Boolean indicating whether or not to print stats.
figName --- Will be used as suptitle and save path, of provided.
Default = None.
saveExt ---
xLabel ---
xLabels ---
yLabel ---
yLim ---
"""
# Check whether 'factor' and 'dv' are column headers in the spread sheet:
vNames = dm.asArray()[0]
if exMask:
dm = dm.select("mask_side == 'control'")
if not factor in vNames:
msg.userMsg.userMsg("The factor %s is not a column header."%factor,\
__file__)
return
if not dv in vNames:
msg.userMsg.userMsg("The dv %s is not a column header."%dv, __file__)
return
driftCorr = dm["driftCorr"][0]
if exControlMask:
dm = dm.select("mask_side != 'control'")
if exFiller:
dm = dm.select("symm == 'asymm'")
# If wanted, trim the data:
if trim:
dm = dm.selectByStdDev(keys = factor + subjectID, dv = dv)
# Get pivot matrix:
pm = PivotMatrix(dm, factor, subjectID, dv=dv, colsWithin=True)
# Get anova matrix:
am = AnovaMatrix(dm, factors = factor, dv = dv, \
subject = subjectID[0])
stats = prettyStats(am)
if printStats:
print am
# Create figure:
fig = plt.figure()
if figName == None:
name = "%s effect on %s - trim = %s exControlMask = %s exFiller = %s driftCorr = %s"%(\
factor, dv, trim, exControlMask, exFiller, driftCorr)
else:
name = figName
plt.suptitle(stats)
# Plot the effect:
#plt.subplot(121)
# Determine some plot properties:
if yLabel == None:
yLabel = dv
if xLabel == None:
xLabel = factor
# Change the labels if one of the factors is 'mask_side':
if factor == ["mask_side"]:
if "control" in np.unique(dm["mask_side"]):
xLabels = ["control", "more contrast right", "more contrast left"]
else:
xLabels = ["more contrast right", "more contrast left"]
pm.barPlot(fig = fig, yLabel = yLabel, xLabel = xLabel, xLabels = xLabels, \
yLim = yLim)
if dv in ["endX", "endXCorr", "endXCorrMask", "endXDegr","endXCorrDegr", "endXCorrMaskDegr"]:
plt.axhline(0, linestyle = "--", color = "#555753")
#statSubplot(am, nRows = 1, nCols = 2, plotNr = 2)
if showFig:
plt.show()
saveFig = False
if saveFig:
plt.savefig(os.path.join(outputFolder,"%s%s"%(name,saveExt)))
src = 'selected_dm_%s_WITH_drift_corr_onlyControl_False.csv' % exp
main_dm = CsvReader(src).dataMatrix()
main_dm = main_dm.select("contrast_side != 'control'")
dv = "endX1NormToContrast"
dm = main_dm.select("%s != ''" % dv)
dm = dm.selectByStdDev(["file"], dv)
dm = dm.removeField("__dummyCond__")
dm = dm.removeField("__stdOutlier__")
dm = dm.selectByStdDev(["file"], "saccLat1")
fig = plt.figure()
pm = PivotMatrix(dm, cols = ["visual_field"], \
rows = ["file"], dv = dv, colsWithin=True)
pm.barPlot(fig = fig)
for vf in ["upper", "lower"]:
_dm = dm.select("visual_field == '%s'" % vf)
cm = _dm.collapse(["file"], dv)
m = cm["mean"].mean()
se = cm['mean'].std() / np.sqrt(len(cm))
print "VISUAL FIELD = ", vf
print "DV = ", dv
print "M = ", m
print "SE = ", se
plt.savefig("VF_effect.png")