def test_split(): dm = DataMatrix(length=4) dm.a = 'a', 'a', 'b', 'b' dm.b = 0, 1, 2, 3 g = ops.split(dm.a) val, dm = g.next() eq_(val, 'a') check_col(dm.a, ['a', 'a']) check_col(dm.b, [0, 1]) val, dm = g.next() eq_(val, 'b') check_col(dm.a, ['b', 'b']) check_col(dm.b, [2, 3])
def test_split():
dm = DataMatrix(length=4)
dm.a = 'a', 'a', 'b', 'b'
dm.b = 0, 1, 2, 3
# Without values
g = ops.split(dm.a)
val, dm = next(g)
assert val == 'a'
check_col(dm.a, ['a', 'a'])
check_col(dm.b, [0, 1])
val, dm = next(g)
assert val == 'b'
check_col(dm.a, ['b', 'b'])
check_col(dm.b, [2, 3])
# With values
dm = DataMatrix(length=4)
dm.a = 'a', 'a', 'b', 'b'
dm.b = 0, 1, 2, 3
dma, dmb = ops.split(dm.a, 'a', 'b')
check_col(dma.a, ['a', 'a'])
check_col(dma.b, [0, 1])
check_col(dmb.a, ['b', 'b'])
check_col(dmb.b, [2, 3])
def roi_hist(dm, col):
for roi, rdm in ops.split(dm.roi):
sns.distplot(rdm[col],
label='{} (N={})'.format(roi, len(rdm)),
hist=True,
kde=False,
bins=50,
hist_kws={
"alpha": .8,
"color": COLORS[roi]
})
plt.xlabel('t value')
plt.ylabel('Density')
plt.axvline(0, color='black', linestyle=':')
plt.legend()
def process_video(run, start_frame=1):
kernel = smoothing_kernel()
cap = cv2.VideoCapture(SRC_VIDEO.format(run - 1))
dm = DataMatrix()
im_prev = None
for frame in it.count(start=start_frame):
ret, im = cap.read()
if not ret or frame >= MAX_FRAME:
print('Done!')
break
dm <<= process_frame(run, frame, luminance_map(im, kernel),
change_map(im, im_prev, kernel))
im_prev = im
for sub, sdm in ops.split(dm.sub):
io.writetxt(sdm, DST.format(subject=sub, run=run))
dm = run()
print(dm)
# Here we set up the baseline to remove NaN values
dm.pupil = srs.endlock(dm.ptrace_rsvp)
dm.pupil = srs.baseline(
series= dm.pupil,
baseline=dm.ptrace_rsvp,
bl_start=0,
bl_end=2,
method='subtractive'
)
# Splitting datamatrix between the two type of stimuli
blueDM, redDM = ops.split(dm.target_object, "blue", "red")
# Plot for preprocessed trace
plt.figure()
plt.title('Preprocesed trace')
for tone, cdm in ops.split(redDM.tone_red):
colour = "#FF3333" if tone == "bright" else "#8B0000"
plot.trace(cdm.pupil, color=colour)
for tone, cdm in ops.split(blueDM.tone_blue):
colour = "#00FFFF" if tone == "bright" else "#00008B"
plot.trace(cdm.pupil, color=colour)
plt.tight_layout()
plt.savefig(path + "/preprocessed.png")
plt.clf()
def test_split():
dm = DataMatrix(length=4)
dm.a = 'a', 'a', 'b', 'b'
dm.b = 0, 1, 2, 3
# Without values
g = ops.split(dm.a)
val, dm = next(g)
assert val == 'a'
check_col(dm.a, ['a', 'a'])
check_col(dm.b, [0, 1])
val, dm = next(g)
assert val == 'b'
check_col(dm.a, ['b', 'b'])
check_col(dm.b, [2, 3])
# With values
dm = DataMatrix(length=4)
dm.a = 'a', 'a', 'b', 'b'
dm.b = 0, 1, 2, 3
dma, dmb = ops.split(dm.a, 'a', 'b')
check_col(dma.a, ['a', 'a'])
check_col(dma.b, [0, 1])
check_col(dmb.a, ['b', 'b'])
check_col(dmb.b, [2, 3])
# With multiple columns
dm = DataMatrix(length=8)
dm.A = 0, 0, 1, 1, 0, 0, 1, 1
dm.B = 'a', 'b', 'a', 'b', 'a', 'b', 'a', 'b'
dm.C = 'x', 'x', 'x', 'x', 'y', 'y', 'y', 'y'
g = ops.split(dm.A, dm.B)
val1, val2, sdm = next(g)
assert val1 == 0
assert val2 == 'a'
assert (len(sdm) == 2)
val1, val2, sdm = next(g)
assert val1 == 0
assert val2 == 'b'
assert (len(sdm) == 2)
val1, val2, sdm = next(g)
assert val1 == 1
assert val2 == 'a'
assert (len(sdm) == 2)
val1, val2, sdm = next(g)
assert val1 == 1
assert val2 == 'b'
assert (len(sdm) == 2)
g = ops.split(dm.A, dm.B, dm.C)
val1, val2, val3, sdm = next(g)
assert val1 == 0
assert val2 == 'a'
assert val3 == 'x'
assert (len(sdm) == 1)
val1, val2, val3, sdm = next(g)
assert val1 == 0
assert val2 == 'a'
assert val3 == 'y'
assert (len(sdm) == 1)
val1, val2, val3, sdm = next(g)
assert val1 == 0
assert val2 == 'b'
assert val3 == 'x'
assert (len(sdm) == 1)
val1, val2, val3, sdm = next(g)
assert val1 == 0
assert val2 == 'b'
assert val3 == 'y'
assert (len(sdm) == 1)
val1, val2, val3, sdm = next(g)
assert val1 == 1
assert val2 == 'a'
assert val3 == 'x'
assert (len(sdm) == 1)
val1, val2, val3, sdm = next(g)
assert val1 == 1
assert val2 == 'a'
assert val3 == 'y'
assert (len(sdm) == 1)
val1, val2, val3, sdm = next(g)
assert val1 == 1
assert val2 == 'b'
assert val3 == 'x'
assert (len(sdm) == 1)
val1, val2, val3, sdm = next(g)
assert val1 == 1
assert val2 == 'b'
assert val3 == 'y'
assert (len(sdm) == 1)
if args.parse:
print("Parsing again! This will take some time...")
run.clear()
else:
if os.path.isdir(path + "/.memoize/"):
print("Using existing parsed data (unless new files have been added).")
else:
print(
"No existing directory with parsed data! Parsing files, this will take some time..."
)
dm = run()
print(dm)
print(dm.column_names)
bwCondition, colourCondition = ops.split(dm.block_condition, 0, 1)
def ttest(var1, var2):
'''Runs a t-test between two variables'''
mean1 = np.mean(var1)
mean2 = np.mean(var2)
vari1 = np.var(var1)
vari2 = np.var(var2)
sd1 = np.sqrt(vari1)
sd2 = np.sqrt(vari2)
len1 = len(var1)
len2 = len(var2)
print("Performing T-Test with the following parameters:")
print("Variable 1 - Mean: {0} SD: {1} N: {2} CV: {3}".format(
mean1, sd1, len1, sd1 / mean1))
def splitFillers(dm):
"""
Splits filler trials into:
- 2 first fillers, one negative, one neutral
- 2 last fillers, one negative, one neutral
- remaining fillers, that should be mixed with the experimental trials
Arguments:
dm --- DataMatrix containing filler objects
Returns:
dm_first --- dm containing only 2 rows, corresponding to the first 2 fillers
dm_last --- dm containing only 2 rows, corresponding to the last 2 fillers
dm_remaining --- dm containing the remainder (if any) of the fillers
NOTE the order
"""
# Below is an awful piece of script to make sure the first and last two fillers
# consist of one neg and one neu trial.
# I just don't see a way to do this more elegantly
# Shuffle the fillers:
dm = ops.shuffle(dm)
# Split the fillers on the basis of valence:
(cat1, filler_dm_negative), (cat2,
filler_dm_neutral) = ops.split(dm["Emotion"])
# Select 2 fillers, one from each valence category, as first 2 fillers:
identifier_filler1_neg = filler_dm_negative["Object"][0]
identifier_filler1_neu = filler_dm_neutral["Object"][0]
dm_filler1_neg = filler_dm_negative["Object"] == identifier_filler1_neg
dm_filler1_neu = filler_dm_neutral["Object"] == identifier_filler1_neu
# Merge the two first trial_dms
dm_first = dm_filler1_neg << dm_filler1_neu
# Shuffle
dm_first = ops.shuffle(dm_first)
# Add a column indicating the type of filler (for later debugging/cross
# checking)
dm_first["filler_type"] = "first"
# Remove the already selected pairs from the dms
# (i.e., select WITHOUT replacement):
filler_dm_negative = filler_dm_negative[1:]
filler_dm_neutral = filler_dm_neutral[1:]
# Do the same thing for the last 2 fillers:
# Select 2 fillers, one from each valence category, as first 2 fillers:
identifier_filler2_neg = filler_dm_negative["Object"][0]
identifier_filler2_neu = filler_dm_neutral["Object"][0]
dm_filler2_neg = filler_dm_negative["Object"] == identifier_filler2_neg
dm_filler2_neu = filler_dm_neutral["Object"] == identifier_filler2_neu
# Merge:
dm_last = dm_filler2_neg << dm_filler2_neu
# And shuffle:
dm_last = ops.shuffle(dm_last)
# Add column with filler type
dm_last["filler_type"] = "last"
# Remove from the dms:
filler_dm_negative = filler_dm_negative[1:]
filler_dm_neutral = filler_dm_neutral[1:]
# Merge the remaining filler dms:
dm_remaining = filler_dm_negative << filler_dm_neutral
# Shuffle:
dm_remaining = ops.shuffle(dm_remaining)
# Add column with filler type
dm_remaining["filler_type"] = "middle"
return dm_first, dm_last, dm_remaining