def test_conditional_ranking(self):
regparam = 0.001
K_train1, K_train2, Y_train, K_test1, K_test2, Y_test, X_train1, X_train2, X_test1, X_test2 = (
self.generate_xortask()
)
train_rows, train_columns = Y_train.shape
trainlabelcount = train_rows * train_columns
K_Kron_train_x = np.kron(K_train2, K_train1)
# Train linear Conditional Ranking Kronecker RLS
params = {}
params["X1"] = X_train1
params["X2"] = X_train2
params["Y"] = Y_train
params["regparam"] = regparam
linear_kron_condrank_learner = KronRLS(**params)
linear_kron_condrank_learner.solve_linear_conditional_ranking(regparam)
# Train an ordinary RankRLS for reference
params = {}
params["X"] = K_Kron_train_x
params["kernel"] = "PrecomputedKernel"
params["Y"] = Y_train.reshape((trainlabelcount, 1), order="F")
qids = []
for j in range(Y_train.shape[1]):
for i in range(Y_train.shape[0]):
qids.append(i)
params["qids"] = qids
rankrls_learner = QueryRankRLS(**params)
rankrls_learner.solve(regparam)
K_test_x = np.kron(K_test2, K_test1)
ordrankrls_testpred = rankrls_learner.predict(K_test_x)
condrank_testpred = linear_kron_condrank_learner.predict(X_test1, X_test2)
# print('')
print(
"\n\nMeanabsdiff: conditional ranking vs rankrls "
+ str(np.mean(np.abs(condrank_testpred - ordrankrls_testpred)))
+ "\n"
)
np.testing.assert_almost_equal(condrank_testpred, ordrankrls_testpred)
def test_conditional_ranking(self):
regparam = 0.001
K_train1, K_train2, Y_train, K_test1, K_test2, Y_test, X_train1, X_train2, X_test1, X_test2 = self.generate_xortask()
train_rows, train_columns = Y_train.shape
trainlabelcount = train_rows * train_columns
K_Kron_train_x = np.kron(K_train2, K_train1)
#Train linear Conditional Ranking Kronecker RLS
params = {}
params["X1"] = X_train1
params["X2"] = X_train2
params["Y"] = Y_train
params["regparam"] = regparam
linear_kron_condrank_learner = KronRLS(**params)
linear_kron_condrank_learner.solve_linear_conditional_ranking(regparam)
#Train an ordinary RankRLS for reference
params = {}
params["X"] = K_Kron_train_x
params["kernel"] = "PrecomputedKernel"
params["Y"] = Y_train.reshape((trainlabelcount, 1), order = 'F')
qids = []
for j in range(Y_train.shape[1]):
for i in range(Y_train.shape[0]):
qids.append(i)
params["qids"] = qids
rankrls_learner = QueryRankRLS(**params)
rankrls_learner.solve(regparam)
K_test_x = np.kron(K_test2, K_test1)
ordrankrls_testpred = rankrls_learner.predict(K_test_x)
condrank_testpred = linear_kron_condrank_learner.predict(X_test1, X_test2)
#print('')
print('\n\nMeanabsdiff: conditional ranking vs rankrls ' + str(np.mean(np.abs(condrank_testpred - ordrankrls_testpred))) + '\n')
np.testing.assert_almost_equal(condrank_testpred, ordrankrls_testpred)
from rlscore.reader import read_qids
from rlscore.reader import read_sparse
from rlscore.reader import read_sparse
from rlscore.reader import read_qids
from rlscore.measure import cindex
train_labels = np.loadtxt("./examples/data/rank_train.labels")
test_labels = np.loadtxt("./examples/data/rank_test.labels")
train_qids = read_qids("./examples/data/rank_train.qids")
test_features = read_sparse("./examples/data/rank_test.features")
train_features = read_sparse("./examples/data/rank_train.features")
test_qids = read_qids("./examples/data/rank_test.qids")
kwargs = {}
kwargs["Y"] = train_labels
kwargs["X"] = train_features
kwargs["qids"] = train_qids
kwargs["regparam"] = 1
learner = QueryRankRLS(**kwargs)
P = learner.predict(test_features)
from rlscore.measure.measure_utilities import UndefinedPerformance
from rlscore.measure.measure_utilities import qids_to_splits
test_qids = qids_to_splits(test_qids)
perfs = []
for query in test_qids:
try:
perf = cindex(test_labels[query], P[query])
perfs.append(perf)
except UndefinedPerformance:
pass
test_perf = np.mean(perfs)
print "test set performance: %f" % test_perf
import numpy as np
from rlscore.learner.query_rankrls import QueryRankRLS
from rlscore.utilities.reader import read_qids
from rlscore.utilities.reader import read_sparse
from rlscore.measure import cindex
train_labels = np.loadtxt("./legacy_tests/data/rank_train.labels")
test_labels = np.loadtxt("./legacy_tests/data/rank_test.labels")
train_qids = read_qids("./legacy_tests/data/rank_train.qids")
test_features = read_sparse("./legacy_tests/data/rank_test.features")
train_features = read_sparse("./legacy_tests/data/rank_train.features")
test_qids = read_qids("./legacy_tests/data/rank_test.qids")
kwargs = {}
kwargs["Y"] = train_labels
kwargs["X"] = train_features
kwargs["qids"] = train_qids
kwargs["regparam"] = 1
learner = QueryRankRLS(**kwargs)
P = learner.predict(test_features)
from rlscore.measure.measure_utilities import UndefinedPerformance
from rlscore.measure.measure_utilities import qids_to_splits
test_qids = qids_to_splits(test_qids)
perfs = []
for query in test_qids:
try:
perf = cindex(test_labels[query], P[query])
perfs.append(perf)
except UndefinedPerformance:
pass
test_perf = np.mean(perfs)
print("test set performance: %f" %test_perf)
