def show():
# load in all the data
participants = list()
studyLengths = list()
for pNum in range(HIGHEST_P_NUMBER + 1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
pid = str(pNum)
try:
lenData = studyLength(fileLoc=settings.getFileName('metaData'))
studyLen = len(lenData.days)
except IOError:
# print 'p '+pid+' metaData file not found'
continue
# if no problems:
participants.append(pid)
studyLengths.append(studyLen)
# print a nice little table of interaction lengths
divider = '======================================'
print divider
cols = ['pid', 'studyLen']
row_format = "{:>15}" * (len(cols))
print row_format.format(*cols)
for pid, length in zip(participants, studyLengths):
print row_format.format(pid, length)
print divider
def show():
# load in all the data
participants = list()
studyLengths = list()
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
pid = str(pNum)
try:
lenData = studyLength(fileLoc=settings.getFileName('metaData'))
studyLen = len(lenData.days)
except IOError:
# print 'p '+pid+' metaData file not found'
continue
# if no problems:
participants.append(pid)
studyLengths.append(studyLen)
# print a nice little table of interaction lengths
divider = '======================================'
print divider
cols = ['pid', 'studyLen']
row_format ="{:>15}" * (len(cols))
print row_format.format(*cols)
for pid, length in zip(participants, studyLengths):
print row_format.format(pid, length)
print divider
def plot():
# load in all the data
participants = list()
interactions = list()
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
try:
pid = str(pNum)
fName = settings.getFileName('viewLog')
interact = (interactionData(fName))
# TODO the rest...
except IOError:
print 'p '+pid+' interactions file not found'
continue
# if no problems:
participants.append(pid)
interactions.append(interact)
print 'p' + pid + ' data loaded.'
#scale all the data
#TODO
print '\n =================================================='
print '\t total participants: '+str(len(participants))
print ' =================================================='
# shift to time since start...
for pnum in range(0,len(interactions)):
start_t = min(interactions[pnum].t)
for i in range (0, len(interactions[pnum])):
interactions[pnum].t[i] -= start_t
# shift all to same daily cycle
for pnum in range(0,len(interactions)):
timeOfDay = interactions[pnum].x[0] #assuming earliest time is first...
ToD_start = (timeOfDay.hour*60 + timeOfDay.minute) * 60
for i in range (0, len(interactions[pnum])):
interactions[pnum].t[i] += ToD_start
# make the plots
figName = "interaction_sparkScatterplot"
pylab.figure(figName)
pylab.subplots_adjust(left=0,bottom=0,right=1,top=1,wspace=0,hspace=0)
num = 0
for interaction in interactions:
pylab.subplot(len(interactions),1,num)
pylab.scatter(interaction.viewTimes,[0]*len(interaction.viewTimes),marker='|',color=interaction.viewMarkerColor)
pylab.plt.gca().axes.get_xaxis().set_visible(False)
num += 1
fname = DATA_LOC+figName+'.png'
pylab.plt.savefig(fname, dpi = 1000)
def plot():
# load in all the data
participants = list()
interactions = list()
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
try:
pid = str(pNum)
fName = settings.getFileName('viewLog')
interact = (interactionData(fName))
except IOError:
print 'p '+pid+' interactions file not found'
continue
# if no problems:
participants.append(pid)
interactions.append(interact)
print 'p' + pid + ' data loaded.'
#scale all the data
#TODO
print '\n =================================================='
print '\t total participants: '+str(len(participants))
print ' =================================================='
# shift to time since start...
for pnum in range(0,len(interactions)):
start_t = min(interactions[pnum].t)
for i in range (0, len(interactions[pnum])):
interactions[pnum].t[i] -= start_t
# shift all to same daily cycle
for pnum in range(0,len(interactions)):
timeOfDay = interactions[pnum].x[0] #assuming earliest time is first...
ToD_start = (timeOfDay.hour*60 + timeOfDay.minute) * 60
for i in range (0, len(interactions[pnum])):
interactions[pnum].t[i] += ToD_start
figName = "interaction_sparkLines"
pylab.figure(figName)
num = 0
for interaction in interactions:
pylab.subplot(len(interactions),1,num)
pylab.plot(interaction.t,interaction.v)
num += 1
fname = DATA_LOC+figName+'.png'
pylab.plt.savefig(fname, dpi = 1000)
def plot():
# load in all the data
participants = list()
PA = list()
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
try:
pid = str(pNum)
fName = settings.getFileName('fitbit')
pa = PAdata(fName,method='fitbit',timeScale='minute')
# TODO the rest...
except IOError:
print 'p '+pid+' PA file "'+ fName +'" not found'
continue
# if no problems:
participants.append(pid)
PA.append(pa)
print 'p' + pid + ' data loaded.'
#scale all the data
#TODO
print '\n =================================================='
print '\t total participants: '+str(len(participants))
print ' =================================================='
# shift to time since start...
# for pnum in range(0,len(PA)):
# start_t = min(PA[pnum].time)
# for i in range (0, len(PA[pnum])):
# PA[pnum].time[i] -= start_t
# # shift all to same daily cycle
# for pnum in range(0,len(PA)):
# timeOfDay = PA[pnum].time[0] #assuming earliest time is first...
# ToD_start = (timeOfDay.hour*60 + timeOfDay.minute) * 60
# for i in range (0, len(PA[pnum])):
# PA[pnum].time[i] += ToD_start
# make the plots
figName = "PA_sparkScatterplot"
pylab.figure(figName)
num = 0
for pa in PA:
pylab.subplot(len(PA),1,num)
pylab.scatter(pa.time, pa.steps, marker='x', color='b')
num += 1
def plot():
# load in all the data
participants = list()
PA = list()
for pNum in range(HIGHEST_P_NUMBER + 1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
try:
pid = str(pNum)
fName = settings.getFileName('fitbit')
pa = PAdata(fName, method='fitbit', timeScale='minute')
# TODO the rest...
except IOError:
print 'p ' + pid + ' PA file "' + fName + '" not found'
continue
# if no problems:
participants.append(pid)
PA.append(pa)
print 'p' + pid + ' data loaded.'
#scale all the data
#TODO
print '\n =================================================='
print '\t total participants: ' + str(len(participants))
print ' =================================================='
# shift to time since start...
# for pnum in range(0,len(PA)):
# start_t = min(PA[pnum].time)
# for i in range (0, len(PA[pnum])):
# PA[pnum].time[i] -= start_t
# # shift all to same daily cycle
# for pnum in range(0,len(PA)):
# timeOfDay = PA[pnum].time[0] #assuming earliest time is first...
# ToD_start = (timeOfDay.hour*60 + timeOfDay.minute) * 60
# for i in range (0, len(PA[pnum])):
# PA[pnum].time[i] += ToD_start
# make the plots
figName = "PA_sparkScatterplot"
pylab.figure(figName)
num = 0
for pa in PA:
pylab.subplot(len(PA), 1, num)
pylab.scatter(pa.time, pa.steps, marker='x', color='b')
num += 1
def plot():
### analysis run on each participant and displayed together ###
# load in all the data
participants = list()
interactions = list()
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
pid = str(pNum)
try:
interact = interactionData(fileName=settings.getFileName('viewLog'),timeScale='daily')
# TODO the rest...
except IOError:
print 'p '+pid+' interactions file not found'
continue
# if no problems:
participants.append(pid)
interactions.append(interact)
print 'p' + pid + ' data loaded.'
#scale all the data
#TODO
print '\n =================================================='
print '\t total participants: '+str(len(participants))
print ' =================================================='
maxStudyLen = max([len(interact.dailyTotals) for interact in interactions])
# make the plots
figName = "interactions_sparkBars"
pylab.figure(figName)
pylab.subplots_adjust(left=0,bottom=0,right=1,top=1,wspace=0,hspace=0)
num = 0
for interact in interactions:
# make all data the same length so bars are sime width
while len(interact.dailyTotals) < maxStudyLen:
interact.dailyTotals.append(-1)
# plot
pylab.subplot(len(interactions),1,num)
pylab.bar(range(len(interact.dailyTotals)),interact.dailyTotals,width=.5)
num += 1
def __init__(self,
settings,
min_quality=QUALITY_LEVEL.acceptable,
used_data_types=DATA_TYPES.all,
trim=True,
check=True,
avatar_view_freq=60):
"""
:param min_quality: minimum quality level of data to be considered in the dataset
"""
self.settings = settings
self.pids = settings.get_pid_list()
self.excluded = settings.get_exluded_list(min_level=min_quality,
used_data=used_data_types)
for pid in list(
self.pids
): # note: need to use copy b/c we are modifying as we go
if pid in self.excluded:
self.pids.remove(pid)
#print 'removing ', pid
set_uses = dict()
for type in used_data_types:
set_uses[type] = True
self.subject_data = list()
for pid in self.pids:
self.subject_data.append(
Subject(setup(dataset=settings.dataset,
data_loc=settings.dataLoc,
subject_n=pid),
avatar_view_freq=avatar_view_freq,
uses=set_uses))
if trim:
self.subject_data[-1].trim_data()
if check:
self.subject_data[-1].integrity_check()
print len(self), 'subjects loaded. pids = ', self.pids
print 'excluding pids ', self.excluded
def get_data(pre_win, post_win):
settings = setup(dataset='test', data_loc='./data/controlIntervention/', subject_n=3)
data = Dataset(
settings,
trim=True,
check=False,
used_data_types=[DATA_TYPES.event, DATA_TYPES.fitbit]
)
minutes = post_win+pre_win
PNUM = 0
bars = []
for evt in data.subject_data[0].event_data.time:
time = evt-timedelta(minutes=pre_win)
bars.append(data.get_steps_after_time(time, minutes, PNUM))
pids = [1]*len(bars) # all events are same participant
return minutes, pids, bars
def get_fake_data(pre_win, post_win, minutes, pids, bars):
# returns data from randomly chosen fake data points
settings = setup(dataset='test', data_loc='./data/controlIntervention/', subject_n=3)
data = Dataset(
settings,
trim=True,
check=False,
used_data_types=[DATA_TYPES.event, DATA_TYPES.fitbit]
)
PNUM = 0
fake_bars = []
for evt in data.subject_data[0].event_data.time:
time = evt-timedelta(days=1, minutes=pre_win) # get random(ish) time
fake_bars.append(data.get_steps_after_time(time, minutes, PNUM))
diff_bars = []
for i in range(len(bars)):
diff_bars.append(list_subtract(bars[i], fake_bars[i]))
return minutes, pids, diff_bars
def plot(dataset='test',dataLoc = "./data/", paMethod=DEFAULT_METHOD, bypass_data_check=False):
# change plot font
# font = {'family' : 'monospace',
# 'weight' : 'normal',
# 'size' : 16}
# pylab.plt.rc('font', **font)
pltName = 'participants\' avg PA for sedentary & active avatar days'
print 'making plot "'+pltName+'"'
pylab.figure(pltName)
cmap = pylab.cm.get_cmap(name='terrain')
activePAs = list()
sedentPAs = list()
zeroPAs = list()
dataCounter = 0 # a count of how many datapoints are used so far (for colormapping and debug)
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
print '===p:'+str(pNum)+'==='
if pNum==1 or pNum==3:
print 'skipping p1 & p3 b/c data is incomplete.\n'
continue
elif pNum==13:
print 'skip p13 b/c data needs some massaging.\n'
continue
elif pNum==14:
print 'skip p14 b/c data is incomplete.\n'
continue
else:
activePAtotal = sedentPAtotal = zeroPAtotal = 0
activePAcount = sedentPAcount = zeroPAcount = 0
try :
settings = setup(dataset=dataset, data_loc=dataLoc, subject_n=pNum)
# load interaction data
interact = interactionData(settings.getFileName('viewLog'))
interactScore,interactDate = segmentInteractionIntoDays(interact)
# load PA data
PA = PAdata(PAfile=settings.getFileName(paMethod), method=paMethod, timeScale='daily')
if paMethod == 'mMonitor':
PAscore, PAdate = segmentPAIntoDays(PA,PAscoreFunction=getPAscore_postiveOnly)
elif paMethod == 'fitbit':
PAscore = PA.steps
PAdate = PA.time
else :
raise ValueError('unknown PA method "'+str(paMethod)+'"')
if not bypass_data_check:
trim_data(interactDate, PAdate, interactScore, PAscore)
interactions = list()
for iScore in interactScore:
if iScore > 0:
interactions.append(1)#'active')
elif iScore < 0:
interactions.append(-1)#'sedentary')
else: # iScore = 0
interactions.append(0)#'0')
for day in range(len(PAscore)):
#print 'day ='+str(day)
#print 'interact='+str(interactions[day])
#print 'PAscore ='+str(PAscore[day])
if interactions[day] > 0:
activePAtotal+=PAscore[day]
activePAcount+=1
elif interactions[day] < 0:
sedentPAtotal+=PAscore[day]
sedentPAcount+=1
else: # interactions[day] == 0
zeroPAtotal+=PAscore[day]
zeroPAcount+=1
activePAs.append(activePAtotal/activePAcount)
sedentPAs.append(sedentPAtotal/sedentPAcount)
try:
zeroPAs.append(zeroPAtotal/zeroPAcount)
except ZeroDivisionError:
zeroPAs.append(0)
# no exceptions means this data was good, increment counter
dataCounter += 1
print str(pNum)+' loaded.\n'
except InputError as e:
print e.message
print 'participant '+str(pNum)+' not valid.\n'
except IOError as e:
print e.message
print 'participant '+str(pNum)+' not found.\n'
except Warning as w:
print w.message
print 'some issues loading participant '+str(pNum)+'; continuing anyway.\n'
if paMethod == 'mMonitor':
pylab.plt.ylabel('physical activity score')
elif paMethod == 'fitbit':
pylab.plt.ylabel('average step counts')
pylab.plt.xlabel('<-sedentary active->\navatar behavior')
pylab.plt.gca().axes.get_xaxis().set_ticks([])
pylab.plt.draw()
print activePAs
print sedentPAs
print zeroPAs
# TODO: colormapping should use dataCounter and actual number of data available (instead of pNum and HIGHEST_P_NUM, respectively)
base = 0
for PA in activePAs:
pylab.plt.bar(1, PA, bottom=base, linewidth=1, width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
base += PA
base = 0
for PA in sedentPAs:
pylab.plt.bar(-1, PA, bottom=base, linewidth=1, width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
base += PA
base = 0
for PA in zeroPAs:
pylab.plt.bar(0, PA, bottom=base, linewidth=1, width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
base += PA
paired_sample = stats.ttest_rel(sedentPAs, activePAs)
print "================================================"
print str(dataCounter) + " subjects analyzed using " + paMethod + " data."
print "The t-statistic is %.3f and the p-value is %.3f." % paired_sample
print "================================================"
print 'done.'
""" analysis run on just one participant: """
import warnings
import pylab
from src.settings import setup
from src.data.subject.Subject import Subject
settings = setup(dataset='USF', data_loc='../subjects/') # use dataset='test' to select sample dataset
with warnings.catch_warnings():
warnings.simplefilter("ignore")
sub = Subject(settings)
sub.trim_data()
sub.integrity_check()
pylab.plt.figure('fitbit time series')
sub.fitbit_data.ts.plot()
pylab.plt.figure('avatar view time series')
sub.avatar_view_data.ts.plot()
pylab.plt.show()
#this one isn't very impressive, and I don't think it is working right now anyway
#import src.interaction.timeSeries.simple
#src.interaction.timeSeries.simple.plot()
#pylab.plt.show()
import src.dep.interaction.timeSeries.multicolorBars
__author__ = 'tylar'
import csv
from src.settings import setup, DATA_TYPES
MIN_LEN = 10 # min legitimate view time in ms
MAX_LEN = 60*1000 # max legitimate view time in ms
REPLACE_TIME = 10*1000 # length of time placed at start and end of illegitimate times
settings = setup(dataset='USF', data_loc='../subjects/', subject_n=0)
old_total = new_total = replaced = removed = kept = 0 # various counters
pids = settings.get_pid_list()
for pid in pids:
view_file_loc = setup(dataset=settings.dataset, data_loc=settings.dataLoc, subject_n=pid).get_file_name(DATA_TYPES.avatar_views)
# read all the rows in
cols = list()
with open(view_file_loc, 'rb') as csvfile:
spamreader = csv.reader(csvfile, delimiter=',')
for row in spamreader:
old_total += 1
#print ', '.join(row) # print the raw data
# print row # print raw data matrix
t0 = int(row[0])
tf = int(row[1])
len = int(row[2])
act = row[3]
def plot(dataset='test',dataLoc = "./data/"):
# change plot font
# font = {'family' : 'monospace',
# 'weight' : 'normal',
# 'size' : 16}
# pylab.plt.rc('font', **font)
pltName = 'All participants\' PA for sedentary & active avatar days'
print 'making plot "'+pltName+'"'
pylab.figure(pltName)
activeBase = 0
sedentBase = 0
zeroBase = 0
cmap = pylab.cm.get_cmap(name='terrain')
activePAs = list()
sedentPAs = list()
zeroPAs = list()
dataCounter = 0 # a count of how many datapoints are used so far (for colormapping and debug)
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
if pNum == 1 or pNum == 3 or pNum == 13 or pNum == 14:
# skip p1 & p3 b/c data is incomplete
# skip p13 b/c data needs some massaging
# skip p14 b/c data is incomplete
continue
try :
settings = setup(dataset=dataset, data_loc=dataLoc, subject_n=pNum)
# load interaction data
interact = interactionData(settings.getFileName('viewLog'))
interactScore,interactDate = segmentInteractionIntoDays(interact)
# load PA data
PA = PAdata(settings.getFileName('mMonitor'))
PAscore,PAdate = segmentPAIntoDays(PA,PAscoreFunction=getPAscore_postiveOnly)
while (interactDate[0].date() != PAdate[0].date()) or (interactDate[-1].date() != PAdate[-1].date()) or (len(interactScore) != len(PAscore)):
print 'day mismatch: '
#print 'data is ' + str(len(PAscore)) + 'x' + str(len(interactScore))
print '\t NAME \tSTART \t\t\tEND \t\t\tLEN'
print '\t interact\t'+str(interactDate[0])+'\t'+str(interactDate[-1])+'\t'+str(len(interactScore))
print '\t PA \t'+str(PAdate[0])+ '\t'+str(PAdate[-1]) +'\t'+str(len(PAscore))+'\n'
if(PAdate[0].date() < interactDate[0].date()):#if pa starts before interact
print 'pa data removed from start'
PAdate.pop(0)
PAscore.pop(0)
elif PAdate[0].date() > interactDate[0].date():#if pa starts after interact
print 'interact data removed from start'
interactDate.pop(0)
interactScore.pop(0)
elif PAdate[-1].date() < interactDate[-1].date() :#if pa ends before interact
print 'pa data removed from end'
interactDate.pop()
interactScore.pop()
elif PAdate[-1].date() > interactDate[-1].date() :#if pa ends after interact
print 'interact data removed from end'
PAscore.pop()
PAdate.pop()
else: # uneven values must be from missing days in middle of one dataset
longer = list()
shorter = list()
ldate = list()
sdate = list()
shortName = ''
if len(PAscore) > len(interactScore):
shortName = 'interaction'
longName = 'PA'
longer = PAscore
shorter= interactScore
ldate = PAdate
sdate = interactDate
else:
shortName = 'PA'
longName = 'interaction'
longer = interactScore
shorter= PAscore
ldate = interactDate
sdate = PAdate
for i in range(len(longer)):
#if i >= len(shorter): #check for end of shortlist reached
# print '\nERR: unknown data mismatch!!!\n'
# print 'data dump:'
# print '=== PA ==='
# print 'DATES='+str(PAdate)
# print 'VALUES='+str(PAscore)
# print '===interact==='
# print 'DATES='+str(interactDate)
# print 'VALUES='+str(interactScore)
# return
if i>=len(shorter) or sdate[i].date() != ldate[i].date():#remove extra dates not in shortlist
print 'value removed from '+longName+' at '+str(ldate[i])
longer.pop(i)
ldate.pop(i)
#shorter.insert(i,0)
#sdate.insert(i,ldate[i])
#print 'zero value inserted into '+shortName+' at '+str(sdate[i])
break
if len(PAscore)<=0 or len(interactScore)<=0:
raise Warning('data has no overlap!')
# modify interaction score to just separate (+) and (-) values
interactions = list()
for iScore in interactScore:
if iScore > 0:
interactions.append(1)#'active')
elif iScore < 0:
interactions.append(-1)#'sedentary')
else: # iScore = 0
interactions.append(0)#'0')
for day in range(len(PAscore)):
#print 'day ='+str(day)
#print 'interact='+str(interactions[day])
#print 'PAscore ='+str(PAscore[day])
if interactions[day] > 0:
activePAs.append(PAscore[day])
# TODO: colormapping should use dataCounter and actual number of data available (instead of pNum and HIGHEST_P_NUM, respectively)
pylab.plt.bar(interactions[day], PAscore[day], bottom=activeBase, linewidth=1, width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
activeBase += PAscore[day]
elif interactions[day] < 0:
sedentPAs.append(PAscore[day])
pylab.plt.bar(interactions[day], PAscore[day], bottom=sedentBase, linewidth=1, width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
sedentBase += PAscore[day]
else: # interactions[day] == 0
zeroPAs.append(PAscore[day])
pylab.plt.bar(interactions[day], PAscore[day], bottom=zeroBase, linewidth=1, width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
zeroBase += PAscore[day]
# no exceptions means this data was good, increment counter
dataCounter += 1
except InputError:
print 'participant '+str(pNum)+' not valid.'
except IOError:
print 'participant '+str(pNum)+' not found.'
except Warning as w:
print w.message
pylab.plt.ylabel('physical activity score')
pylab.plt.xlabel('<-sedentary zero active->\navatar behavior')
pylab.plt.draw()
print activePAs
print sedentPAs
print zeroPAs
paired_sample = stats.ttest_ind(sedentPAs, activePAs)
print "================================================"
print "The t-statistic is %.3f and the p-value is %.3f." % paired_sample
print "================================================"
print 'done.'
""" analysis run on just one participant: """
import warnings
import pylab
from src.settings import setup
from src.data.subject.Subject import Subject
settings = setup(
dataset='USF',
data_loc='../subjects/') # use dataset='test' to select sample dataset
with warnings.catch_warnings():
warnings.simplefilter("ignore")
sub = Subject(settings)
sub.trim_data()
sub.integrity_check()
pylab.plt.figure('fitbit time series')
sub.fitbit_data.ts.plot()
pylab.plt.figure('avatar view time series')
sub.avatar_view_data.ts.plot()
pylab.plt.show()
#this one isn't very impressive, and I don't think it is working right now anyway
#import src.interaction.timeSeries.simple
#src.interaction.timeSeries.simple.plot()
#pylab.plt.show()
def plot():
### analysis run on each participant and displayed together ###
# load in all the data
participants = list()
interactions = list()
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
pid = str(pNum)
try:
interact = (interactionData(settings.getFileName('viewLog')))
# TODO the rest...
except IOError:
print 'p '+pid+' interactions file not found'
continue
# if no problems:
participants.append(pid)
interactions.append(interact)
print 'p' + pid + ' data loaded.'
#scale all the data
#TODO
print '\n =================================================='
print '\t total participants: '+str(len(participants))
print ' =================================================='
# figName = "all_interactions_x_time"
# pylab.figure(figName)
# pylab.plot(interactions[0].x,interactions[0].v,
# interactions[1].x,interactions[1].v,
# interactions[2].x,interactions[2].v,
# interactions[3].x,interactions[3].v,
# interactions[4].x,interactions[4].v,
# interactions[5].x,interactions[5].v,
# interactions[6].x,interactions[6].v,
# interactions[7].x,interactions[7].v,
# interactions[8].x,interactions[8].v,
# interactions[9].x,interactions[9].v,
# interactions[10].x,interactions[10].v) #TODO: add more if applicable
# fname = DATA_LOC+figName+'.png'
# pylab.plt.savefig(fname, dpi = 100)
# shift to time since start...
for pnum in range(0,len(interactions)):
start_t = min(interactions[pnum].t)
for i in range (0, len(interactions[pnum])):
interactions[pnum].t[i] -= start_t
# figName = "all_interactions_x_time_since_study_start"
# pylab.figure(figName)
# pylab.plot(interactions[0].t,interactions[0].v,
# interactions[1].t,interactions[1].v,
# interactions[2].t,interactions[2].v,
# interactions[3].t,interactions[3].v,
# interactions[4].t,interactions[4].v,
# interactions[5].t,interactions[5].v,
# interactions[6].t,interactions[6].v,
# interactions[7].t,interactions[7].v,
# interactions[8].t,interactions[8].v,
# interactions[9].t,interactions[9].v,
# interactions[10].t,interactions[10].v) #TODO: add more if applicable
# fname = DATA_LOC+figName+'.png'
# pylab.plt.savefig(fname, dpi = 100)
# shift all to same daily cycle
for pnum in range(0,len(interactions)):
timeOfDay = interactions[pnum].x[0] #assuming earliest time is first...
ToD_start = (timeOfDay.hour*60 + timeOfDay.minute) * 60
for i in range (0, len(interactions[pnum])):
interactions[pnum].t[i] += ToD_start
figName = "all_interactions_x_time_of_day_since_study_start"
pylab.figure(figName)
try:
assert(len(interactions) == 15)
except AssertionError as e:
print 'len(interactions)=',len(interactions)
raise
pylab.plot(interactions[0].t,interactions[0].v,
interactions[1].t,interactions[1].v,
interactions[2].t,interactions[2].v,
interactions[3].t,interactions[3].v,
interactions[4].t,interactions[4].v,
interactions[5].t,interactions[5].v,
interactions[6].t,interactions[6].v,
interactions[7].t,interactions[7].v,
interactions[8].t,interactions[8].v,
interactions[9].t,interactions[9].v,
interactions[10].t,interactions[10].v,
interactions[11].t,interactions[11].v,
interactions[12].t,interactions[12].v,
interactions[13].t,interactions[13].v,
interactions[14].t,interactions[14].v) #TODO: add more if applicable
fname = DATA_LOC+figName+'.png'
pylab.plt.savefig(fname, dpi = 100)
import knowMe
# these are switches to help debugging, set true to run on each part
avatar = False
knowme = True
# some optional tests:
# from src.post_view_event_steps_bars import test_get_avg_list
# test_get_avg_list()
#knowMe.makePlots(type=PLOT_TYPES.bars, show=True, pre_win=10, post_win=40)
#knowMe.makePlots(type=PLOT_TYPES.bars, show=True)
if avatar:
### USF mAVATAR DATA LOADING ###
settings = setup(dataset='USF', data_loc='../subjects/', subject_n=0)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
data = Dataset(settings, min_quality=QUALITY_LEVEL.acceptable, trim=True, check=True,
used_data_types=[DATA_TYPES.fitbit, DATA_TYPES.avatar_views], avatar_view_freq=60)
UP_TO_DATE = True # true if software versions are good
if pandas.version.version < '0.12.0':
UP_TO_DATE = False
print '\n\nWARN: Some analysis cannot be completed due to outdated pandas version ' + pandas.version.version + '\n\n'
# comparison of events selected with/without overlap from mAvatar dataset
# to demonstrate difference (especially at high time intervals like no-overlap for 3hrs around event)
import warnings
import pylab
import pandas
from src.settings import setup, QUALITY_LEVEL, DATA_TYPES
from src.data.mAvatar.Data import DAY_TYPE
from src.data.Dataset import Dataset
settings = setup(dataset='USF', data_loc='../subjects/', subject_n=0)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
data = Dataset(
settings,
min_quality=QUALITY_LEVEL.acceptable,
trim=True,
check=True,
used_data_types=[DATA_TYPES.fitbit, DATA_TYPES.avatar_views],
avatar_view_freq=60)
UP_TO_DATE = True # true if software versions are good
if pandas.version.version < '0.12.0':
UP_TO_DATE = False
print '\n\nWARN: Some analysis cannot be completed due to outdated pandas version ' + pandas.version.version + '\n\n'
###################
### BEGIN plots ###
###################
if UP_TO_DATE:
def plot():
# load in all the data
PA_participants = list()
PA = list()
for pNum in range(HIGHEST_P_NUMBER + 1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
try:
pid = pNum
fName = settings.getFileName('fitbit')
pa = PAdata(fName, method='fitbit', timeScale='minute')
except IOError:
print 'p ' + str(pid) + ' PA file "' + fName + '" not found'
continue
# if no problems:
PA_participants.append(pid)
PA.append(pa)
# print 'p' + pid + ' data loaded.'
interact_participants = list()
interactions = list()
for pNum in range(HIGHEST_P_NUMBER + 1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
try:
pid = pNum
fName = settings.getFileName('viewLog')
interact = (interactionData(fName))
except IOError:
print 'p ' + str(pid) + ' interactions file not found'
continue
# if no problems:
interact_participants.append(pid)
interactions.append(interact)
# print 'p' + pid + ' data loaded.'
#scale all the data
PAscaler = 1.0 / 50.0
for pnum in range(0, len(PA)):
for i in range(0, len(PA[pnum])):
PA[pnum].steps[i] *= PAscaler
# shift to time since start...
# for pnum in range(0,len(PA)):
# start_t = min(PA[pnum].timestamp)
# for i in range (0, len(PA[pnum])):
# PA[pnum].timestamp[i] -= start_t
# for pnum in range(0,len(interactions)):
# start_t = min(interactions[pnum].t)
# for i in range (0, len(interactions[pnum])):
# interactions[pnum].t[i] -= start_t
# # shift all to same daily cycle
# for pnum in range(0,len(PA)):
# timeOfDay = PA[pnum].time[0] #assuming earliest time is first...
# for i in range (0, len(PA[pnum])):
# PA[pnum].time[i] += timeOfDay
# timeOfDay = interactions[pnum].t[0] #assuming earliest time is first...
# ToD_start = (timeOfDay.hour*60 + timeOfDay.minute) * 60
# for i in range (0, len(interactions[pnum])):
# interactions[pnum].t[i] += ToD_start
# class Bunch:
# ''' bunches! makes a dictionary look like a class! bam! '''
# __init__ = lambda self, **kw: setattr(self, '__dict__', kw)
# make the plots
figName = "scatterplot_masterdash"
pylab.figure(figName)
pylab.subplots_adjust(left=0, bottom=0, right=1, top=1, wspace=0, hspace=0)
num = 0
max_p = max(max(PA_participants), max(interact_participants)) + 1
pCount = max(len(PA_participants), len(interact_participants)) + 1
actualCount = 1
for i in range(max_p):
pylab.subplot(pCount, 1, actualCount)
try:
pa = PA[PA_participants.index(num)]
except ValueError as m:
print 'p' + m.message + ' of PA'
pa = None # Bunch(time=0,steps=0)
try:
interact = interactions[interact_participants.index(num)]
except ValueError as m:
print 'p' + m.message + ' of interactions'
interact = None # Bunch(t=0,v=0)
num += 1
isData = False
try:
pylab.scatter(pa.timestamp, pa.steps, marker='x', color='g')
isData = True
except AttributeError as e:
pass
try:
pylab.scatter(interact.viewTimes, [0] * len(interact.viewTimes),
marker='+',
color=interact.viewMarkerColor)
isData = True
except AttributeError as e:
pass
if isData:
actualCount += 1
def plot():
### analysis run on each participant and displayed together ###
# load in all the data
participants = list()
interactions = list()
for pNum in range(HIGHEST_P_NUMBER + 1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
pid = str(pNum)
try:
interact = (interactionData(settings.getFileName('viewLog')))
# TODO the rest...
except IOError:
print 'p ' + pid + ' interactions file not found'
continue
# if no problems:
participants.append(pid)
interactions.append(interact)
print 'p' + pid + ' data loaded.'
#scale all the data
#TODO
print '\n =================================================='
print '\t total participants: ' + str(len(participants))
print ' =================================================='
# figName = "all_interactions_x_time"
# pylab.figure(figName)
# pylab.plot(interactions[0].x,interactions[0].v,
# interactions[1].x,interactions[1].v,
# interactions[2].x,interactions[2].v,
# interactions[3].x,interactions[3].v,
# interactions[4].x,interactions[4].v,
# interactions[5].x,interactions[5].v,
# interactions[6].x,interactions[6].v,
# interactions[7].x,interactions[7].v,
# interactions[8].x,interactions[8].v,
# interactions[9].x,interactions[9].v,
# interactions[10].x,interactions[10].v) #TODO: add more if applicable
# fname = DATA_LOC+figName+'.png'
# pylab.plt.savefig(fname, dpi = 100)
# shift to time since start...
for pnum in range(0, len(interactions)):
start_t = min(interactions[pnum].t)
for i in range(0, len(interactions[pnum])):
interactions[pnum].t[i] -= start_t
# figName = "all_interactions_x_time_since_study_start"
# pylab.figure(figName)
# pylab.plot(interactions[0].t,interactions[0].v,
# interactions[1].t,interactions[1].v,
# interactions[2].t,interactions[2].v,
# interactions[3].t,interactions[3].v,
# interactions[4].t,interactions[4].v,
# interactions[5].t,interactions[5].v,
# interactions[6].t,interactions[6].v,
# interactions[7].t,interactions[7].v,
# interactions[8].t,interactions[8].v,
# interactions[9].t,interactions[9].v,
# interactions[10].t,interactions[10].v) #TODO: add more if applicable
# fname = DATA_LOC+figName+'.png'
# pylab.plt.savefig(fname, dpi = 100)
# shift all to same daily cycle
for pnum in range(0, len(interactions)):
timeOfDay = interactions[pnum].x[
0] #assuming earliest time is first...
ToD_start = (timeOfDay.hour * 60 + timeOfDay.minute) * 60
for i in range(0, len(interactions[pnum])):
interactions[pnum].t[i] += ToD_start
figName = "all_interactions_x_time_of_day_since_study_start"
pylab.figure(figName)
try:
assert (len(interactions) == 15)
except AssertionError as e:
print 'len(interactions)=', len(interactions)
raise
pylab.plot(interactions[0].t, interactions[0].v, interactions[1].t,
interactions[1].v, interactions[2].t, interactions[2].v,
interactions[3].t, interactions[3].v, interactions[4].t,
interactions[4].v, interactions[5].t, interactions[5].v,
interactions[6].t, interactions[6].v, interactions[7].t,
interactions[7].v, interactions[8].t, interactions[8].v,
interactions[9].t, interactions[9].v, interactions[10].t,
interactions[10].v, interactions[11].t, interactions[11].v,
interactions[12].t, interactions[12].v, interactions[13].t,
interactions[13].v, interactions[14].t,
interactions[14].v) #TODO: add more if applicable
fname = DATA_LOC + figName + '.png'
pylab.plt.savefig(fname, dpi=100)
def plot():
# load in all the data
PA_participants = list()
PA = list()
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
try:
pid = pNum
fName = settings.getFileName('fitbit')
pa = PAdata(fName,method='fitbit',timeScale='minute')
except IOError:
print 'p '+str(pid)+' PA file "'+ fName +'" not found'
continue
# if no problems:
PA_participants.append(pid)
PA.append(pa)
# print 'p' + pid + ' data loaded.'
interact_participants = list()
interactions = list()
for pNum in range(HIGHEST_P_NUMBER+1): #cycle through all participants
settings = setup(dataset='USF', data_loc=DATA_LOC, subject_n=pNum)
try:
pid = pNum
fName = settings.getFileName('viewLog')
interact = (interactionData(fName))
except IOError:
print 'p '+str(pid)+' interactions file not found'
continue
# if no problems:
interact_participants.append(pid)
interactions.append(interact)
# print 'p' + pid + ' data loaded.'
#scale all the data
PAscaler = 1.0/50.0
for pnum in range (0,len(PA)):
for i in range(0,len(PA[pnum])):
PA[pnum].steps[i] *= PAscaler
# shift to time since start...
# for pnum in range(0,len(PA)):
# start_t = min(PA[pnum].timestamp)
# for i in range (0, len(PA[pnum])):
# PA[pnum].timestamp[i] -= start_t
# for pnum in range(0,len(interactions)):
# start_t = min(interactions[pnum].t)
# for i in range (0, len(interactions[pnum])):
# interactions[pnum].t[i] -= start_t
# # shift all to same daily cycle
# for pnum in range(0,len(PA)):
# timeOfDay = PA[pnum].time[0] #assuming earliest time is first...
# for i in range (0, len(PA[pnum])):
# PA[pnum].time[i] += timeOfDay
# timeOfDay = interactions[pnum].t[0] #assuming earliest time is first...
# ToD_start = (timeOfDay.hour*60 + timeOfDay.minute) * 60
# for i in range (0, len(interactions[pnum])):
# interactions[pnum].t[i] += ToD_start
# class Bunch:
# ''' bunches! makes a dictionary look like a class! bam! '''
# __init__ = lambda self, **kw: setattr(self, '__dict__', kw)
# make the plots
figName = "scatterplot_masterdash"
pylab.figure(figName)
pylab.subplots_adjust(left=0,bottom=0,right=1,top=1,wspace=0,hspace=0)
num = 0
max_p = max(max(PA_participants),max(interact_participants))+1
pCount = max(len(PA_participants),len(interact_participants))+1
actualCount = 1
for i in range(max_p):
pylab.subplot(pCount,1,actualCount)
try:
pa = PA[PA_participants.index(num)]
except ValueError as m:
print 'p' + m.message + ' of PA'
pa = None # Bunch(time=0,steps=0)
try:
interact = interactions[interact_participants.index(num)]
except ValueError as m:
print 'p' + m.message + ' of interactions'
interact = None # Bunch(t=0,v=0)
num += 1
isData = False
try:
pylab.scatter(pa.timestamp, pa.steps, marker='x', color = 'g')
isData = True
except AttributeError as e:
pass
try:
pylab.scatter(interact.viewTimes, [0]*len(interact.viewTimes), marker='+', color=interact.viewMarkerColor)
isData = True
except AttributeError as e:
pass
if isData:
actualCount += 1
def setUp(self):
from src.settings import setup
self.settings = setup(dataset='default')
def plot(dataset='test',
dataLoc="./data/",
paMethod=DEFAULT_METHOD,
bypass_data_check=False):
# change plot font
# font = {'family' : 'monospace',
# 'weight' : 'normal',
# 'size' : 16}
# pylab.plt.rc('font', **font)
pltName = 'participants\' avg PA for sedentary & active avatar days'
print 'making plot "' + pltName + '"'
pylab.figure(pltName)
cmap = pylab.cm.get_cmap(name='terrain')
activePAs = list()
sedentPAs = list()
zeroPAs = list()
dataCounter = 0 # a count of how many datapoints are used so far (for colormapping and debug)
for pNum in range(HIGHEST_P_NUMBER + 1): #cycle through all participants
print '===p:' + str(pNum) + '==='
if pNum == 1 or pNum == 3:
print 'skipping p1 & p3 b/c data is incomplete.\n'
continue
elif pNum == 13:
print 'skip p13 b/c data needs some massaging.\n'
continue
elif pNum == 14:
print 'skip p14 b/c data is incomplete.\n'
continue
else:
activePAtotal = sedentPAtotal = zeroPAtotal = 0
activePAcount = sedentPAcount = zeroPAcount = 0
try:
settings = setup(dataset=dataset,
data_loc=dataLoc,
subject_n=pNum)
# load interaction data
interact = interactionData(settings.getFileName('viewLog'))
interactScore, interactDate = segmentInteractionIntoDays(
interact)
# load PA data
PA = PAdata(PAfile=settings.getFileName(paMethod),
method=paMethod,
timeScale='daily')
if paMethod == 'mMonitor':
PAscore, PAdate = segmentPAIntoDays(
PA, PAscoreFunction=getPAscore_postiveOnly)
elif paMethod == 'fitbit':
PAscore = PA.steps
PAdate = PA.time
else:
raise ValueError('unknown PA method "' + str(paMethod) +
'"')
if not bypass_data_check:
trim_data(interactDate, PAdate, interactScore, PAscore)
interactions = list()
for iScore in interactScore:
if iScore > 0:
interactions.append(1) #'active')
elif iScore < 0:
interactions.append(-1) #'sedentary')
else: # iScore = 0
interactions.append(0) #'0')
for day in range(len(PAscore)):
#print 'day ='+str(day)
#print 'interact='+str(interactions[day])
#print 'PAscore ='+str(PAscore[day])
if interactions[day] > 0:
activePAtotal += PAscore[day]
activePAcount += 1
elif interactions[day] < 0:
sedentPAtotal += PAscore[day]
sedentPAcount += 1
else: # interactions[day] == 0
zeroPAtotal += PAscore[day]
zeroPAcount += 1
activePAs.append(activePAtotal / activePAcount)
sedentPAs.append(sedentPAtotal / sedentPAcount)
try:
zeroPAs.append(zeroPAtotal / zeroPAcount)
except ZeroDivisionError:
zeroPAs.append(0)
# no exceptions means this data was good, increment counter
dataCounter += 1
print str(pNum) + ' loaded.\n'
except InputError as e:
print e.message
print 'participant ' + str(pNum) + ' not valid.\n'
except IOError as e:
print e.message
print 'participant ' + str(pNum) + ' not found.\n'
except Warning as w:
print w.message
print 'some issues loading participant ' + str(
pNum) + '; continuing anyway.\n'
if paMethod == 'mMonitor':
pylab.plt.ylabel('physical activity score')
elif paMethod == 'fitbit':
pylab.plt.ylabel('average step counts')
pylab.plt.xlabel(
'<-sedentary active->\navatar behavior'
)
pylab.plt.gca().axes.get_xaxis().set_ticks([])
pylab.plt.draw()
print activePAs
print sedentPAs
print zeroPAs
# TODO: colormapping should use dataCounter and actual number of data available (instead of pNum and HIGHEST_P_NUM, respectively)
base = 0
for PA in activePAs:
pylab.plt.bar(
1, PA, bottom=base, linewidth=1,
width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
base += PA
base = 0
for PA in sedentPAs:
pylab.plt.bar(
-1, PA, bottom=base, linewidth=1,
width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
base += PA
base = 0
for PA in zeroPAs:
pylab.plt.bar(
0, PA, bottom=base, linewidth=1,
width=1.7) #, color=cmap(float(pNum)/float(HIGHEST_P_NUMBER)) )
base += PA
paired_sample = stats.ttest_rel(sedentPAs, activePAs)
print "================================================"
print str(dataCounter) + " subjects analyzed using " + paMethod + " data."
print "The t-statistic is %.3f and the p-value is %.3f." % paired_sample
print "================================================"
print 'done.'
def setUp(self): from src.settings import setup self.settings = setup(dataset='default')
