def train_test_NN(train,
labels,
test,
use_rescale_priors=False,
normalize_log=True,
extra_feature_count=0,
extra_feature_seed=0,
**parameters):
"""
Train and test a neural network given a set of parameters (which should contain no iterables). Returns test data probabilities for use in (parallel) optimizer.
"""
net, train, test = train_NN(train,
labels,
test,
use_rescale_priors=use_rescale_priors,
normalize_log=normalize_log,
extra_feature_count=extra_feature_count,
extra_feature_seed=extra_feature_seed,
**parameters)
prediction = net.predict_proba(test)
if use_rescale_priors:
prediction = scale_to_priors(prediction,
priors=bincount(labels)[1:] /
float64(len(labels)))
return prediction
def predict(parameters, networkfile, data):
"""
Calculate probabilities for the data.
:param parameters: parameters for the network (must match networkfile)
:param networkfile: .net.npz file for the network
:param data: ndarray with data (train or test)
:return: probabilities
Doesn't work with added test data or with outlier removal.
"""
parameters = copy(parameters)
parameters.update({'verbosity': False, 'pretrain': networkfile})
net = train_NN(data, labels = None, test = None, test_only = True, **parameters)[0]
load_knowledge(net, networkfile)
prediction = net.predict_proba(data)
scale_to_priors(prediction, priors = normalized_sum([1929, 16122, 8004, 2691, 2739, 14135, 2839, 8464, 4955]))
print 'predicted {0:d} samples'.format(prediction.shape[0])
return prediction
def train_test(train, labels, test, n_neighbors, distance_p, use_log = False, use_autoscale = False, use_calibration = False): if use_log: train, test = log10(1 + train), log10(1 + test) if use_autoscale: train /= train.max(0) test /= test.max(0) clf = DistanceClassifier(n_neighbors = n_neighbors, distance_p = distance_p) if use_calibration: clf = CalibratedClassifierCV(clf, cv = 3) clf.fit(train, labels) probs = clf.predict_proba(test) probs = scale_to_priors(probs) return probs
def train_test_NN(train,
labels,
test,
use_rescale_priors=False,
outlier_frac=0,
outlier_method='OCSVM',
normalize_log=True,
use_calibration=False,
**parameters):
net = make_net(**parameters)
train, test = conormalize_data(train, test, use_log=normalize_log)
load_knowledge(net, 'results/nnets/optimize_new.log_1000.net.npz')
prediction = net.predict_proba(test)
if use_rescale_priors:
prediction = scale_to_priors(prediction,
priors=bincount(labels)[1:] /
float64(len(labels)))
return prediction
names = ['final_final1_351', 'final_final2_5019', 'final_final1_4969', 'final_final2_5530', 'final_final2_2247', 'final_final1_8594', 'final_final1_1717', 'final_final4_3641', 'final_final2_9535', 'final_final2_2066', 'final_final2_5878', 'final_final4_7076', 'final_final3_6441', 'final_final3_5475']
totalvalid = totaltest = 0
for name in names:
net = NNet.load(filepath = join(BASE_DIR, 'results', 'nets', name))
for nm, val in net.get_params().iteritems():
print '{0:s} = {1:}'.format(nm, val)
#for nm, data in [('val', valid), ('tst', test)]:
#probs = net.predict_proba(data)
#save(join(SUBMISSIONS_DIR, '{0}_{1}_raw.npy'.format(name, nm)), probs)
#makeSubmission(probs, fname = join(SUBMISSIONS_DIR, '{0}_{1}_rescale.csv'.format(name, nm)), digits = 8)
probs = net.predict_proba(valid)
probs = scale_to_priors(probs, priors = PRIORS)
save(join(SUBMISSIONS_DIR, '{0}_valid.npy'.format(name)), probs)
totalvalid += probs
probs = net.predict_proba(test)
probs = scale_to_priors(probs, priors = PRIORS)
save(join(SUBMISSIONS_DIR, '{0}_test.npy'.format(name)), probs)
makeSubmission(probs, fname = join(SUBMISSIONS_DIR, '{0}_test.csv'.format(name)), digits = 8)
totaltest += probs
save(join(SUBMISSIONS_DIR, 'total_valid.npy'), totalvalid)
save(join(SUBMISSIONS_DIR, 'total_valid.npy'), totaltest)
makeSubmission(totaltest, fname = join(SUBMISSIONS_DIR, 'total_test.csv'), digits = 8)
print 'saved predictions'
if not isfile(pretrain):
print '>> pretraining network'
make_pretrain(pretrain, train, labels, extra_feature_count = extra_feature_count, **params)
print '>> loading pretrained network'
load_knowledge(net, pretrain)
print '>> training network'
out = net.fit(train, labels - 1)
print '>> saving network'
save_knowledge(net, join(NNET_STATE_DIR, 'single_trained.net.npz'))
print '>> calculating train error'
prediction = net.predict_proba(train)
prediction = scale_to_priors(prediction, priors = bincount(labels)[1:] / float64(len(labels)))
print 'train loss: {0:.4f} / {0:.4f} (unscaled / scaled)'.format(calc_logloss(prediction, labels))
print '>> predicting test data'
prediction = net.predict_proba(test)
print '>> scaling to priors'
prediction = scale_to_priors(prediction, priors = bincount(labels)[1:] / float64(len(labels)))
print '>> making submission file'
make_submission(prediction, fname = join(SUBMISSIONS_DIR, 'single.csv'), digits = 8)
print '>> plotting training progress'
fig, ax = show_train_progress(net)
print '>> done!'