def main():
global thresholdIG
#builds the data vectors for both datasets..
#liFeatures,trainVectors, testVectors = buildDataVectors(DataSet.Three, FeatureSelection.DocFrequency)
#liFeatures,trainVectors, testVectors = buildDataVectors(DataSet.Two, FeatureSelection.InformationGain)
#print testVectors[0], len(testVectors)
#logisticRegression.runLogisticRegression(liFeatures, trainVectors, testVectors)
handleYelpDataset()
exit()
numTries = 10
logResults = {} #dictionary to a tuple of (runtime in seconds, accuracy)
naiveResults = {}
randomResults = {}
liThresholds=[]
sum1 = 0
for i in range(numTries):
liThresholds.append(sum1)
sum1 += 0.05 #remove 5% extra terms each time..
#print sum1
#lids = [DataSet.One, DataSet.Two]
lids = [DataSet.Two]
liLenFeatures = []
for ds in lids:
if ds == DataSet.One:
print "="*40
print "Using data set one."
print "="*40
else:
print "="*40
print "Using data set two."
print "="*40
for k in range(numTries): #total 10 different number of features..
thresholdIG = liThresholds[k]
liFeatures,trainVectors, testVectors = buildDataVectors(ds, FeatureSelection.InformationGain)
liLenFeatures.append(len(liFeatures))
lAccuracy = 0
#perform logistic regression..
start_time = time.time()
lAccuracy = logisticRegression.runLogisticRegression(liFeatures, trainVectors, testVectors)
logResults[len(liFeatures)] = (time.time() - start_time, lAccuracy)
#perform naive Bayes...
start_time = time.time()
lAccuracy = naiveBayes.runNaiveBayes(liFeatures, trainVectors, testVectors)
naiveResults[len(liFeatures)] = (time.time() - start_time, lAccuracy)
#perform random forest...
#TODO
liLenFeatures.sort()
print "-"*40
print "Printing results with Logistic Regression"
print "-"*40
print "(numFeatures)\t(accuracy)\t(runTime)"
for k in liLenFeatures:
print k, "\t", logResults[k][1], "\t", logResults[k][0]
print "-"*40
print "Printing results with Naive Bayes"
print "-"*40
print "(numFeatures)\t(accuracy)\t(runTime)"
for k in liLenFeatures:
print k, "\t", naiveResults[k][1], "\t", naiveResults[k][0]
'''
def main():
global thresholdIG
#builds the data vectors for both datasets..
#liFeatures,trainVectors, testVectors = buildDataVectors(DataSet.Three, FeatureSelection.DocFrequency)
#liFeatures,trainVectors, testVectors = buildDataVectors(DataSet.Two, FeatureSelection.InformationGain)
#print testVectors[0], len(testVectors)
#logisticRegression.runLogisticRegression(liFeatures, trainVectors, testVectors)
handleYelpDataset()
exit()
numTries = 10
logResults = {} #dictionary to a tuple of (runtime in seconds, accuracy)
naiveResults = {}
randomResults = {}
liThresholds = []
sum1 = 0
for i in range(numTries):
liThresholds.append(sum1)
sum1 += 0.05 #remove 5% extra terms each time..
#print sum1
#lids = [DataSet.One, DataSet.Two]
lids = [DataSet.Two]
liLenFeatures = []
for ds in lids:
if ds == DataSet.One:
print "=" * 40
print "Using data set one."
print "=" * 40
else:
print "=" * 40
print "Using data set two."
print "=" * 40
for k in range(numTries): #total 10 different number of features..
thresholdIG = liThresholds[k]
liFeatures, trainVectors, testVectors = buildDataVectors(
ds, FeatureSelection.InformationGain)
liLenFeatures.append(len(liFeatures))
lAccuracy = 0
#perform logistic regression..
start_time = time.time()
lAccuracy = logisticRegression.runLogisticRegression(
liFeatures, trainVectors, testVectors)
logResults[len(liFeatures)] = (time.time() - start_time, lAccuracy)
#perform naive Bayes...
start_time = time.time()
lAccuracy = naiveBayes.runNaiveBayes(liFeatures, trainVectors,
testVectors)
naiveResults[len(liFeatures)] = (time.time() - start_time,
lAccuracy)
#perform random forest...
#TODO
liLenFeatures.sort()
print "-" * 40
print "Printing results with Logistic Regression"
print "-" * 40
print "(numFeatures)\t(accuracy)\t(runTime)"
for k in liLenFeatures:
print k, "\t", logResults[k][1], "\t", logResults[k][0]
print "-" * 40
print "Printing results with Naive Bayes"
print "-" * 40
print "(numFeatures)\t(accuracy)\t(runTime)"
for k in liLenFeatures:
print k, "\t", naiveResults[k][1], "\t", naiveResults[k][0]
'''
def handleYelpDataset(): liFeatures,trainVectors, testVectors = buildDataVectors(DataSet.Yelp, FeatureSelection.InformationGain) print "running log reg now.." logisticRegression.runLogisticRegression(liFeatures, trainVectors, testVectors)
def handleYelpDataset():
liFeatures, trainVectors, testVectors = buildDataVectors(
DataSet.Yelp, FeatureSelection.InformationGain)
print "running log reg now.."
logisticRegression.runLogisticRegression(liFeatures, trainVectors,
testVectors)
