if len(sys.argv) != 4:
print >> sys.stderr, "Usage: classify.py [TAB_FILE] [classifier: tree, bayes, svm, logreg] [number to filter, 0 -> no filtering]"
sys.exit(1)
data = proj_utils.load_data(sys.argv[1])
type = sys.argv[2]
features = int(sys.argv[3])
train_data, test_data = proj_utils.partition_data(data)
model = train_classifier(train_data, type, features)
train_CA, train_results = proj_utils.test_classifier(model, train_data)
test_CA, test_results = proj_utils.test_classifier(model, test_data)
#print "Train Accuracy: %f, Test Accuracy: %f" % (train_CA, test_CA)
train_stats = proj_utils.get_stats(train_results)
test_stats = proj_utils.get_stats(test_results)
print "Train:\n%s" % str(train_stats)
print "\nTest:\n%s" % str(test_stats)
if features != 0:
print "Features selected:\n%s" % str(model.domain)
filename = '%s%sresults%s%s_filtered_%s.txt' % (
os.path.dirname(__file__), os.sep, os.sep, type, str(features)
) #+ type + '_filtered_' + str(features) + '.txt'
else:
filename = '%s%sresults%s%s.txt' % (os.path.dirname(__file__), os.sep,
os.sep, type)
print filename
sys.exit(1)
proportions = 0.1
if len(sys.argv) >= 3:
proportions = float(sys.argv[2])
data = proj_utils.load_data(sys.argv[1])
train_data, test_data = proj_utils.partition_data(data)
print "\"Proportion\"",
proj_utils.print_csv_header()
#for prop in (0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0):
for prop in (1.0, 1.5, 2.0, 2.5, 3.0, 3.5):
resampled_train_data = proj_utils.undersample_class(
train_data, 'nonad', prop)
model = train_classifier(resampled_train_data)
train_CA, train_results = proj_utils.test_classifier(
model, resampled_train_data)
test_CA, test_results = proj_utils.test_classifier(model, test_data)
#print "Train Accuracy: %f, Test Accuracy: %f" % (train_CA, test_CA)
#train_stats = proj_utils.get_stats(train_results)
test_stats = proj_utils.get_stats(test_results)
#print "Train:\n%s" % str(train_stats)
#print "\nTest:\n%s\n" % str(test_stats)
print "%f, " % prop,
proj_utils.print_results_csv(test_stats)
if len(sys.argv) != 4:
print >> sys.stderr, "Usage: classify.py [TAB_FILE] [classifier: tree, bayes, svm, logreg] [number to filter, 0 -> no filtering]"
sys.exit(1)
data = proj_utils.load_data(sys.argv[1])
type = sys.argv[2]
features = int(sys.argv[3])
train_data, test_data = proj_utils.partition_data(data)
model = train_classifier(train_data, type, features)
train_CA, train_results = proj_utils.test_classifier(model, train_data)
test_CA, test_results = proj_utils.test_classifier(model, test_data)
#print "Train Accuracy: %f, Test Accuracy: %f" % (train_CA, test_CA)
train_stats = proj_utils.get_stats(train_results)
test_stats = proj_utils.get_stats(test_results)
print "Train:\n%s" % str(train_stats)
print "\nTest:\n%s" % str(test_stats)
if features != 0:
print "Features selected:\n%s" % str(model.domain)
filename = '%s%sresults%s%s_filtered_%s.txt' % (os.path.dirname(__file__), os.sep, os.sep, type, str(features)) #+ type + '_filtered_' + str(features) + '.txt'
else:
filename = '%s%sresults%s%s.txt' % (os.path.dirname(__file__), os.sep, os.sep, type)
print filename
f = open(filename, 'w+')
f.write("Train:\n")
f.write(str(train_stats) + "\n")
sys.exit(1)
proportions = 0.1
if len(sys.argv) >= 3:
proportions = float(sys.argv[2])
data = proj_utils.load_data(sys.argv[1])
train_data, test_data = proj_utils.partition_data(data)
print "\"Proportion\"",
proj_utils.print_csv_header()
#for prop in (0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0):
for prop in (1.0,1.5,2.0,2.5,3.0,3.5):
resampled_train_data = proj_utils.undersample_class(train_data, 'nonad', prop)
model = train_classifier(resampled_train_data)
train_CA, train_results = proj_utils.test_classifier(model, resampled_train_data)
test_CA, test_results = proj_utils.test_classifier(model, test_data)
#print "Train Accuracy: %f, Test Accuracy: %f" % (train_CA, test_CA)
#train_stats = proj_utils.get_stats(train_results)
test_stats = proj_utils.get_stats(test_results)
#print "Train:\n%s" % str(train_stats)
#print "\nTest:\n%s" % str(test_stats)
print "%f, " % prop,
proj_utils.print_results_csv(test_stats)