def fitTreeOnTrials(self, trialList, maxLength=-1, person='global'):
#Handles fasttrees ifan implementation. Definition of the fasttrees class was only adjusted to become compitable with up-to-date python and numpy versions
if FFTtool.fc != None:
return
for item in trialList:
for a in self.componentKeys:
if a not in item.keys():
continue
if item['conservatism'] >= 3.5:
if 'Republicans' in a:
item[a.replace('Republicans', 'Party')] = item[a]
item.pop(a, None)
item.pop(a.replace('Republicans', 'Democrats'))
elif item['conservatism'] <= 3.5:
if 'Democrats' in a:
item[a.replace('Democrats', 'Party')] = item[a]
item.pop(a, None)
item.pop(a.replace('Democrats', 'Republicans'))
if 'Sent' in a:
if a.split(' ')[1] not in SentimentAnalyzer.relevant:
continue
item[a] = SentimentAnalyzer.analysis(
item['item'])[a.split(' ')[1]]
data = self.toDFFormatTruthful(trialList)
FFTtool.fc = FastFrugalTreeClassifier(max_levels=5)
FFTtool.fc.fit(data.drop(columns='response'), data['response'])
def predictS(self, item, **kwargs):
#prepare item features format and partisanship
if len(kwargs.keys()) == 1:
kwargs = kwargs['kwargs']
try:
if 'aux' not in item.keys():
item['aux'] = item
except:
tempitem = item
item = {}
item['item'] = tempitem
item['aux'] = kwargs
for a in FFTmax.componentKeys:
if 'Sent' in a:
if a.split(' ')[1] not in SentimentAnalyzer.relevant:
continue
item['aux'][a] = SentimentAnalyzer.analysis(item['item'])[a.split(' ')[1]]
if a.replace('Republicans', 'Party') not in item['aux'].keys() and a.replace('Democrats', 'Party') not in item['aux'].keys():
continue
if item['aux']['conservatism'] >= 3.5:
if 'Republicans' in a:
item['aux'][a.replace('Republicans', 'Party')] = item['aux'][a]
item['aux'].pop(a,None)
item['aux'].pop(a.replace('Republicans','Democrats'))
elif item['aux']['conservatism'] <= 3.5:
if 'Democrats' in a:
item['aux'][a.replace('Democrats', 'Party')] = item['aux'][a]
item['aux'].pop(a,None)
item['aux'].pop(a.replace('Democrats','Republicans'))
#evaluate FFT from root node on
return FFTtool.MAX.run(item, **kwargs, show=False)
def predictS(self, item, **kwargs):
if len(kwargs.keys()) == 1:
kwargs = kwargs['kwargs']
analysis = SentimentAnalyzer.analysis(item)
p = 0
for a in self.parameter.keys():
if a.split(' ')[1] not in self.relevant:
continue
p += analysis[a]* self.parameter[a]
return 1 if self.thresh < p else 0
def predictS(self, item, person='global', **kwargs):
if 'item' not in kwargs.keys():
kwargs['item'] = item
for a in self.componentKeys:
if kwargs['conservatism'] >= 3.5:
if 'Republicans' in a and a in kwargs.keys():
kwargs[a.replace('Republicans', 'Party')] = kwargs[a]
elif kwargs['conservatism'] <= 3.5:
if 'Democrats' in a and a in kwargs.keys():
kwargs[a.replace('Democrats', 'Party')] = kwargs[a]
if 'Sent' in a:
kwargs[a] = SentimentAnalyzer.analysis(
kwargs['item'])[a.split(' ')[1]]
pred = FFTtool.fc.predict(self.toDFFormat([kwargs]))
return int(pred[0])
def fitTreeOnTrials(self, trialList, maxLength=-1, person='global'):
if FFTtool.MAX != None:
return
for item in trialList:
for a in FFTmax.componentKeys:
if a not in item.keys():
continue
if item['conservatism'] >= 3.5:
if 'Republicans' in a:
item[a.replace('Republicans', 'Party')] = item[a]
item.pop(a,None)
item.pop(a.replace('Republicans','Democrats'))
elif item['conservatism'] <= 3.5:
if 'Democrats' in a:
item[a.replace('Democrats', 'Party')] = item[a]
item.pop(a,None)
item.pop(a.replace('Democrats','Republicans'))
if 'Sent' in a:
if a.split(' ')[1] not in SentimentAnalyzer.relevant:
continue
item[a] = SentimentAnalyzer.analysis(item['item'])[a.split(' ')[1]]
maxLength = -1
predictionQuality = {}
predictionMargin = {}
for a in FFTmax.componentKeys:
a = a.replace('Democrats','Party')
a = a.replace('Republicans','Party')
if '<' + a in predictionMargin.keys():
continue
#calculate predictive quality of individual cues
marginOptimum = basinhopping(parametrizedPredictiveQualityLT, [0.00], niter=60, stepsize=3.0, T=.9, minimizer_kwargs={"args" : (a,trialList), "tol":0.001, "bounds" : [[0,5]]},disp=0)
predictionMargin['>' + a] = marginOptimum.x[0]
predictionQuality['>' + a] = marginOptimum.fun
marginOptimum = basinhopping(parametrizedPredictiveQualityST, [0.00], niter=60, stepsize=3.0, T=.9, minimizer_kwargs={"args" : (a,trialList), "tol":0.001, "bounds" : [[0,5]]},disp=0)
predictionMargin['<' + a] = marginOptimum.x[0]
predictionQuality['<' + a] = marginOptimum.fun
#calculate order and direction of cues
orderedConditions = []
for a in sorted(predictionQuality.items(), key=lambda x: x[1], reverse=False):
if a[0][1:] not in item.keys():
continue
if a[0][1:] not in [i[1:] for i in orderedConditions] and a[0][1:] in FFTmax.componentKeys:
orderedConditions.append(a[0])
#assemble tree
for sa in orderedConditions[:maxLength] if maxLength > 0 else orderedConditions:
b = sa[1:]
s = sa[0]
cond = 'item[\'aux\'][\'' + b + '\'] ' + s + ' ' + str(predictionMargin[sa])
newnode = Node(cond,True,False)
rep0preds, rep1preds, length0, length1 = predictiveQuality(newnode, trialList)
if self.fft == None:
self.fft = newnode
self.lastnode = self.fft
else:
if rep1preds/length1 >= rep0preds/length0:
self.lastnode.left = newnode
self.lastnode = self.lastnode.left
elif rep1preds/length1 <= rep0preds/length0:
self.lastnode.right = newnode
self.lastnode = self.lastnode.right
FFTtool.MAX = self.fft
def fitTreeOnTrials(self, trialList, maxLength=-1, person='global'):
for item in trialList:
for a in FFTzigzag.componentKeys:
if 'Sent' in a:
if a.split(' ')[1] not in SentimentAnalyzer.relevant:
continue
item[a] = SentimentAnalyzer.analysis(item['item'])[a.split(' ')[1]]
if a not in item.keys():
continue
if item['conservatism'] >= 3.5:
if 'Republicans' in a:
item[a.replace('Republicans', 'Party')] = item[a]
item.pop(a,None)
item.pop(a.replace('Republicans','Democrats'))
elif item['conservatism'] <= 3.5:
if 'Democrats' in a:
item[a.replace('Democrats', 'Party')] = item[a]
item.pop(a,None)
item.pop(a.replace('Democrats','Republicans'))
maxLength = -1
predictionQuality = {}
predictionMargin = {}
for a in FFTzigzag.componentKeys:
a = a.replace('Democrats','Party')
a = a.replace('Republicans','Party')
if '<' + a in predictionMargin.keys():
continue
if item['conservatism'] >= 3.5:
if 'Republicans' in a:
a = a.replace('Republicans','Party')
elif item['conservatism'] <= 3.5:
if 'Democrats' in a:
a = a.replace('Democrats', 'Party')
#calculate predictive quality of individual cues
marginOptimum = basinhopping(parametrizedPredictiveQualityLT, [0.00], niter=60, stepsize=3.0, T=.9, minimizer_kwargs={"args" : (a,trialList), "tol":0.001, "bounds" : [[0,5]]},disp=0)
predictionMargin['>' + a] = marginOptimum.x[0]
predictionQuality['>' + a] = marginOptimum.fun
marginOptimum = basinhopping(parametrizedPredictiveQualityST, [0.00], niter=60, stepsize=3.0, T=.9, minimizer_kwargs={"args" : (a,trialList), "tol":0.001, "bounds" : [[0,5]]},disp=0)
predictionMargin['<' + a] = marginOptimum.x[0]
predictionQuality['<' + a] = marginOptimum.fun
orderedConditionsPos = []
orderedConditionsNeg = []
#calculate order and direction of cues for both Accept (Pos) and Reject (Neg) exits
for a in sorted(predictionQuality.items(), key=lambda x: x[1], reverse=False):
if a[0][1:] not in item.keys():
continue
b = a[0][1:]
s = a[0][0]
cond = 'item[\'aux\'][\'' + b + '\'] ' + s + ' ' + str(predictionMargin[a[0]])
newnode = Node(cond,True,False)
rep0preds, rep1preds, length0, length1 = predictiveQuality(newnode, trialList)
#determine exit direction
if rep1preds/length1 >= rep0preds/length0:
if a[0][1:] not in [i[1:] for i in orderedConditionsPos + orderedConditionsNeg] and a[0][1:] in [a.replace('Democrats','Party').replace('Republicans','Party') for a in FFTzigzag.componentKeys]:
orderedConditionsPos.append(a[0])
else:
if a[0][1:] not in [i[1:] for i in orderedConditionsNeg +orderedConditionsPos] and a[0][1:] in [a.replace('Democrats','Party').replace('Republicans','Party') for a in FFTzigzag.componentKeys]:
orderedConditionsNeg.append(a[0])
orderedConditions = []
for i in range(max(len(orderedConditionsNeg), len(orderedConditionsPos))):
if len(orderedConditionsNeg) > i:
orderedConditions.append(orderedConditionsNeg[i])
if len(orderedConditionsPos) > i:
orderedConditions.append(orderedConditionsPos[i])
exitLeft = True #for first exit, as Z+ version implemented
#assemble tree
for sa in orderedConditions[:maxLength] if maxLength > 0 else orderedConditions:
b = sa[1:]
s = sa[0]
cond = 'item[\'aux\'][\'' + b + '\'] ' + s + ' ' + str(predictionMargin[sa])
newnode = Node(cond,True,False)
if self.fft == None:
self.fft = newnode
self.lastnode = self.fft
else:
if not exitLeft:
self.lastnode.left = newnode
self.lastnode = self.lastnode.left
else:
self.lastnode.right = newnode
self.lastnode = self.lastnode.right
exitLeft = not exitLeft
FFTtool.ZigZag = self.fft
def fitTreeOnTrials(self, trialList, maxLength=-1, person='global'):
for item in trialList:
for a in FFTzigzag.componentKeys:
if 'Sent' in a:
if a.split(' ')[1] not in SentimentAnalyzer.relevant:
continue
item[a] = SentimentAnalyzer.analysis(item['item'])[a.split(' ')[1]]
if a not in item.keys():
continue
if item['conservatism'] >= 3.5:
if 'Republicans' in a:
item[a.replace('Republicans', 'Party')] = item[a]
item.pop(a,None)
item.pop(a.replace('Republicans','Democrats'))
elif item['conservatism'] <= 3.5:
if 'Democrats' in a:
item[a.replace('Democrats', 'Party')] = item[a]
item.pop(a,None)
item.pop(a.replace('Democrats','Republicans'))
maxLength = -1
predictionQuality = {}
predictionMargin = {}
#precalculated predictive quality of individual cues
predictionQuality, predictionMargin = {'>crt': -0.9996858309770656, '<crt': -0.9986962190352021, '>conservatism': -0.5007337607167683, '<conservatism': -0.5003962614779187, '>reaction_time': -0.6742837053963789, '<reaction_time': -0.5003962614779187, '>Familiarity_Party_Combined': -0.9999108416547788, '<Familiarity_Party_Combined': -0.5003962614779187, '>Partisanship_All_Combined': -0.9996858309770656, '<Partisanship_All_Combined': -0.9986962190352021, '>Sent: negative_emotion': -0.9996849401386263, '<Sent: negative_emotion': -0.5003962614779187, '>Sent: health': -0.5003962614779187, '<Sent: health': -0.6130589430894309, '>Sent: dispute': -0.6450213266656861, '<Sent: dispute': -0.5003962614779187, '>Sent: government': -0.9986893840104849, '<Sent: government': -0.5003962614779187, '>Sent: healing': -0.9993434011818779, '<Sent: healing': -0.5003962614779187, '>Sent: military': -0.9998210130660462, '<Sent: military': -0.5003962614779187, '>Sent: fight': -0.7681743973878805, '<Sent: fight': -0.5003962614779187, '>Sent: meeting': -0.9997459349593496, '<Sent: meeting': -0.5003962614779187, '>Sent: shape_and_size': -0.9996849401386263, '<Sent: shape_and_size': -0.5003962614779187, '>Sent: power': -0.9998427425695864, '<Sent: power': -0.5003962614779187, '>Sent: terrorism': -0.9996849401386263, '<Sent: terrorism': -0.5003962614779187, '>Sent: competing': -0.9996849401386263, '<Sent: competing': -0.5003962614779187, '>Sent: office': -0.9996846420687481, '<Sent: office': -0.5003962614779187, '>Sent: money': -0.999787007454739, '<Sent: money': -0.5003962614779187, '>Sent: aggression': -0.9999039385206532, '<Sent: aggression': -0.5003962614779187, '>Sent: wealthy': -0.9997457411645054, '<Sent: wealthy': -0.5003962614779187, '>Sent: banking': -0.9993421052631579, '<Sent: banking': -0.5003962614779187, '>Sent: kill': -0.9998209169054442, '<Sent: kill': -0.5003962614779187, '>Sent: business': -0.9998209169054442, '<Sent: business': -0.5003962614779187, '>Sent: speaking': -0.999842668344871, '<Sent: speaking': -0.5003962614779187, '>Sent: work': -0.9998209169054442, '<Sent: work': -0.5003962614779187, '>Sent: valuable': -0.9995617879053462, '<Sent: valuable': -0.5003962614779187, '>Sent: economics': -0.9998593134496342, '<Sent: economics': -0.5003962614779187, '>Sent: payment': -0.9993421052631579, '<Sent: payment': -0.5003962614779187, '>Sent: friends': -0.9986807387862797, '<Sent: friends': -0.5003962614779187, '>Sent: giving': -0.999344262295082, '<Sent: giving': -0.5003962614779187, '>Sent: help': -0.9986893840104849, '<Sent: help': -0.7597607052896725, '>Sent: school': -0.5003962614779187, '<Sent: school': -0.7597607052896725, '>Sent: college': -0.5003962614779187, '<Sent: college': -0.7597607052896725, '>Sent: real_estate': -0.9996851385390428, '<Sent: real_estate': -0.5003962614779187, '>Sent: reading': -0.9986893840104849, '<Sent: reading': -0.5003962614779187, '>Sent: gain': -0.9996851385390428, '<Sent: gain': -0.5003962614779187, '>Sent: science': -0.9996851385390428, '<Sent: science': -0.5003962614779187, '>Sent: negotiate': -0.9986893840104849, '<Sent: negotiate': -0.5003962614779187, '>Sent: law': -0.9998427920138343, '<Sent: law': -0.5003962614779187, '>Sent: crime': -0.9998213966779782, '<Sent: crime': -0.5003962614779187, '>Sent: stealing': -0.9996861268047709, '<Sent: stealing': -0.5003962614779187, '>Sent: strength': -0.9997928319867413, '<Sent: strength': -0.5003962614779187}, {'>crt': 2.108692258820941, '<crt': 4.221334524079497, '>conservatism': 2.4733894476129064, '<conservatism': 0.0, '>reaction_time': 2.0723292361934, '<reaction_time': 0.0, '>Familiarity_Party_Combined': 2.1152623939462805, '<Familiarity_Party_Combined': 0.0, '>Partisanship_All_Combined': 2.1077281741718545, '<Partisanship_All_Combined': 4.307574112948173, '>Sent: negative_emotion': 0.0, '<Sent: negative_emotion': 0.0, '>Sent: health': 2.6557183446611052, '<Sent: health': 9.278284718691123e-06, '>Sent: dispute': 0.0, '<Sent: dispute': 0.0, '>Sent: government': 1.951016698230048, '<Sent: government': 0.0, '>Sent: healing': 0.0, '<Sent: healing': 0.0, '>Sent: military': 0.0, '<Sent: military': 0.0, '>Sent: fight': 0.0, '<Sent: fight': 0.0, '>Sent: meeting': 0.0, '<Sent: meeting': 0.0, '>Sent: shape_and_size': 0.0, '<Sent: shape_and_size': 0.0, '>Sent: power': 0.0, '<Sent: power': 0.0, '>Sent: terrorism': 0.0, '<Sent: terrorism': 0.0, '>Sent: competing': 0.0, '<Sent: competing': 0.0, '>Sent: office': 0.0, '<Sent: office': 0.0, '>Sent: money': 0.0, '<Sent: money': 0.0, '>Sent: aggression': 0.0, '<Sent: aggression': 0.0, '>Sent: wealthy': 0.0, '<Sent: wealthy': 0.0, '>Sent: banking': 0.0, '<Sent: banking': 0.0, '>Sent: kill': 0.0, '<Sent: kill': 0.0, '>Sent: business': 0.0, '<Sent: business': 0.0, '>Sent: speaking': 0.0, '<Sent: speaking': 0.0, '>Sent: work': 0.0, '<Sent: work': 0.0, '>Sent: valuable': 0.0, '<Sent: valuable': 0.0, '>Sent: economics': 0.0, '<Sent: economics': 0.0, '>Sent: payment': 0.0, '<Sent: payment': 0.0, '>Sent: friends': 0.0, '<Sent: friends': 0.0, '>Sent: giving': 1.303250036513945, '<Sent: giving': 0.0, '>Sent: help': 1.3873013971092425, '<Sent: help': 9.278284717033553e-06, '>Sent: school': 2.0401223021398005, '<Sent: school': 9.278284719291338e-06, '>Sent: college': 1.4173002482924701, '<Sent: college': 9.278284719291338e-06, '>Sent: real_estate': 0.0, '<Sent: real_estate': 0.0, '>Sent: reading': 0.0, '<Sent: reading': 0.0, '>Sent: gain': 0.0, '<Sent: gain': 0.0, '>Sent: science': 0.0, '<Sent: science': 0.0, '>Sent: negotiate': 1.7659083823077841, '<Sent: negotiate': 0.0, '>Sent: law': 0.0, '<Sent: law': 0.0, '>Sent: crime': 0.0, '<Sent: crime': 0.0, '>Sent: stealing': 0.0, '<Sent: stealing': 0.0, '>Sent: strength': 0.0, '<Sent: strength': 0.0}
orderedConditionsPos = []
orderedConditionsNeg = []
#calculate order and direction of cues for both Accept (Pos) and Reject (Neg) exits
for a in sorted(predictionQuality.items(), key=lambda x: x[1], reverse=False):
if a[0][1:] not in item.keys():
continue
b = a[0][1:]
s = a[0][0]
cond = 'item[\'aux\'][\'' + b + '\'] ' + s + ' ' + str(predictionMargin[a[0]])
newnode = Node(cond,True,False)
rep0preds, rep1preds, length0, length1 = predictiveQuality(newnode, trialList)
#determine exit direction
if rep1preds/length1 >= rep0preds/length0:
if a[0][1:] not in [i[1:] for i in orderedConditionsPos + orderedConditionsNeg] and a[0][1:] in [a.replace('Democrats','Party').replace('Republicans','Party') for a in FFTzigzag.componentKeys]:
orderedConditionsPos.append(a[0])
else:
if a[0][1:] not in [i[1:] for i in orderedConditionsNeg +orderedConditionsPos] and a[0][1:] in [a.replace('Democrats','Party').replace('Republicans','Party') for a in FFTzigzag.componentKeys]:
orderedConditionsNeg.append(a[0])
orderedConditions = []
for i in range(max(len(orderedConditionsNeg), len(orderedConditionsPos))):
if len(orderedConditionsNeg) > i:
orderedConditions.append(orderedConditionsNeg[i])
if len(orderedConditionsPos) > i:
orderedConditions.append(orderedConditionsPos[i])
exitLeft = True #for first exit, as Z+ version implemented
#assemble tree
for sa in orderedConditions[:maxLength] if maxLength > 0 else orderedConditions:
b = sa[1:]
s = sa[0]
cond = 'item[\'aux\'][\'' + b + '\'] ' + s + ' ' + str(predictionMargin[sa])
newnode = Node(cond,True,False)
if self.fft == None:
self.fft = newnode
self.lastnode = self.fft
else:
if not exitLeft:
self.lastnode.left = newnode
self.lastnode = self.lastnode.left
else:
self.lastnode.right = newnode
self.lastnode = self.lastnode.right
exitLeft = not exitLeft
FFTtool.ZigZag = self.fft
def fitTreeOnTrials(self, trialList, maxLength=-1, person='global'):
for item in trialList:
for a in FFTzigzag.componentKeys:
if 'Sent' in a:
if a.split(' ')[1] not in SentimentAnalyzer.relevant:
continue
item[a] = SentimentAnalyzer.analysis(
item['item'])[a.split(' ')[1]]
if a not in item.keys():
continue
if item['conservatism'] >= 3.5:
if 'Republicans' in a:
item[a.replace('Republicans', 'Party')] = item[a]
item.pop(a, None)
item.pop(a.replace('Republicans', 'Democrats'))
elif item['conservatism'] <= 3.5:
if 'Democrats' in a:
item[a.replace('Democrats', 'Party')] = item[a]
item.pop(a, None)
item.pop(a.replace('Democrats', 'Republicans'))
maxLength = -1
predictionQuality = {}
predictionMargin = {}
#precalculated predictive quality of individual cues
predictionQuality, predictionMargin = {
'>crt': -0.499969808586438,
'<crt': -0.499969808586438,
'>conservatism': -0.499969808586438,
'<conservatism': -0.499969808586438,
'>ct': -0.499969808586438,
'<ct': -0.499969808586438,
'>education': -0.499969808586438,
'<education': -0.499969808586438,
'>accimp': -0.499969808586438,
'<accimp': -0.499969808586438,
'>panasPos': -0.499969808586438,
'<panasPos': -0.499969808586438,
'>panasNeg': -0.499969808586438,
'<panasNeg': -0.499969808586438,
'>Exciting_Party_Combined': -0.499969808586438,
'<Exciting_Party_Combined': -0.9967320261437909,
'>Familiarity_Party_Combined': -0.9996378123868164,
'<Familiarity_Party_Combined': -0.499969808586438,
'>Importance_Party_Combined': -0.9978308026030369,
'<Importance_Party_Combined': -0.7963446475195822,
'>Partisanship_All_Combined': -0.9978308026030369,
'<Partisanship_All_Combined': -0.777589954117363,
'>Partisanship_All_Partisan': -0.9978308026030369,
'<Partisanship_All_Partisan': -0.9997585124366095,
'>Partisanship_Party_Combined': -0.9967845659163987,
'<Partisanship_Party_Combined': -0.7990093847758082,
'>Worrying_Party_Combined': -0.499969808586438,
'<Worrying_Party_Combined': -0.9935897435897436
}, {
'>crt': 1.972872196401318,
'<crt': 0.0,
'>conservatism': 0.0,
'<conservatism': 0.0,
'>ct': 0.0,
'<ct': 0.0,
'>education': 0.0,
'<education': 0.0,
'>accimp': 0.0,
'<accimp': 0.0,
'>panasPos': 0.0,
'<panasPos': 0.0,
'>panasNeg': 0.0,
'<panasNeg': 0.0,
'>Exciting_Party_Combined': 0.0,
'<Exciting_Party_Combined': 3.574226008657847,
'>Familiarity_Party_Combined': 2.6042025215277933,
'<Familiarity_Party_Combined': 0.0,
'>Importance_Party_Combined': 2.2036394876853738,
'<Importance_Party_Combined': 4.255574086579121,
'>Partisanship_All_Combined': 1.965200689848336,
'<Partisanship_All_Combined': 3.8353730503393244,
'>Partisanship_All_Partisan': 1.2298451980672356,
'<Partisanship_All_Partisan': 0.7176463353940941,
'>Partisanship_Party_Combined': 4.223723273224007,
'<Partisanship_Party_Combined': 3.885513240139143,
'>Worrying_Party_Combined': 0.0,
'<Worrying_Party_Combined': 1.6378138233962547
}
orderedConditionsPos = []
orderedConditionsNeg = []
#calculate order and direction of cues for both Accept (Pos) and Reject (Neg) exits
for a in sorted(predictionQuality.items(),
key=lambda x: x[1],
reverse=False):
if a[0][1:] not in item.keys():
continue
b = a[0][1:]
s = a[0][0]
cond = 'item[\'aux\'][\'' + b + '\'] ' + s + ' ' + str(
predictionMargin[a[0]])
newnode = Node(cond, True, False)
rep0preds, rep1preds, length0, length1 = predictiveQuality(
newnode, trialList)
#determine exit direction
if rep1preds / length1 >= rep0preds / length0:
if a[0][1:] not in [
i[1:]
for i in orderedConditionsPos + orderedConditionsNeg
] and a[0][1:] in [
a.replace('Democrats', 'Party').replace(
'Republicans', 'Party')
for a in FFTzigzag.componentKeys
]:
orderedConditionsPos.append(a[0])
else:
if a[0][1:] not in [
i[1:]
for i in orderedConditionsNeg + orderedConditionsPos
] and a[0][1:] in [
a.replace('Democrats', 'Party').replace(
'Republicans', 'Party')
for a in FFTzigzag.componentKeys
]:
orderedConditionsNeg.append(a[0])
orderedConditions = []
for i in range(
max(len(orderedConditionsNeg), len(orderedConditionsPos))):
if len(orderedConditionsNeg) > i:
orderedConditions.append(orderedConditionsNeg[i])
if len(orderedConditionsPos) > i:
orderedConditions.append(orderedConditionsPos[i])
exitLeft = True #for first exit, as Z+ version implemented
#assemble tree
for sa in orderedConditions[:
maxLength] if maxLength > 0 else orderedConditions:
b = sa[1:]
s = sa[0]
cond = 'item[\'aux\'][\'' + b + '\'] ' + s + ' ' + str(
predictionMargin[sa])
newnode = Node(cond, True, False)
if self.fft == None:
self.fft = newnode
self.lastnode = self.fft
else:
if not exitLeft:
self.lastnode.left = newnode
self.lastnode = self.lastnode.left
else:
self.lastnode.right = newnode
self.lastnode = self.lastnode.right
exitLeft = not exitLeft
FFTtool.ZigZag = self.fft
